Octopus and Barracuda - Dual-Core Aegir Installer with 10+ Platforms - deploy, upgrade and clone Aegir instances in 8 minutes! (big update)

Posted by omega8cc on August 26, 2010 at 5:05pm

UPDATE: Barracuda and Octopus new home is on d.o, plus there is BOA group:

http://drupal.org/project/barracuda
http://drupal.org/project/octopus
http://groups.drupal.org/boa

NOTE: this thread is a continuation of previously single-core Barracuda Installer introduction: http://groups.drupal.org/node/84074.

Instead of any standard long explanation, I will simply post here the install task log with some pictures. Remember, it is 100% done on command line, just by running the new Octopus Installer. Aegir front-end was not used to install it all.

http://skitch.com/omega8cc/du3u1/fullscreen-2
http://skitch.com/omega8cc/du3u8/fullscreen

BTW. Drupal 7.0-alpha6 works on the latest Aegir 0.4-alpha12, see:

http://skitch.com/omega8cc/du3wm/fullscreen
http://skitch.com/omega8cc/du3w3/fullscreen

Both Octopus and Barracuda Installers are available from our GitHub repository: http://github.com/omega8cc/nginx-for-drupal (read README.txt and check docs to learn more).

Test, submit issues in the GitHub queue, share and enjoy!

v246a:~# bash OCTOPUS.sh.txt
NEW AEGIR setup in progress..
_DISTRO is 001
RAW _DOMAIN is host.aegir.cc
 
==> START -> checkpoint: 

  * Your e-mail address appears to be noc@omega8.cc - is that correct?
  * Your Aegir control panel for this instance will be available at http://host.aegir.cc.
  * Your Aegir user for this instance will be host.
    
 
Do you want to proceed with the setup? [Y/n] Y
==> Thu Aug 26 16:08:59 UTC 2010 INIT A: Welcome Linux v246a.us.host8.biz
==> Thu Aug 26 16:09:08 UTC 2010 INIT A: Aegir automated install script part A
==> Thu Aug 26 16:09:08 UTC 2010 INIT A: Stop the cron and wait 8 seconds..
==> Thu Aug 26 16:09:18 UTC 2010 INIT A: Checking status..
==> Thu Aug 26 16:09:27 UTC 2010 INIT A: _STATUS is INIT
==> Thu Aug 26 16:09:27 UTC 2010 INIT A: Create directories and/or set good permissions
==> Thu Aug 26 16:09:36 UTC 2010 INIT A: Add user
==> Thu Aug 26 16:09:46 UTC 2010 INIT A: Prepare scripts
==> Thu Aug 26 16:09:55 UTC 2010 INIT A: Run standard pre-installer
==> Thu Aug 26 16:10:13 UTC 2010 INIT A: Run standard pre-installer completed
==> Thu Aug 26 16:10:13 UTC 2010 INIT A: Switch user and run AegirSetupB
==> Thu Aug 26 16:10:22 UTC 2010 INIT B: Aegir automated install script part B
==> Thu Aug 26 16:10:22 UTC 2010 INIT B: Checking status..
==> Thu Aug 26 16:10:31 UTC 2010 INIT B: Create dirs and/or set good permissions
==> Thu Aug 26 16:10:40 UTC 2010 INIT B: NEW AEGIR setup in progress..
==> Thu Aug 26 16:10:40 UTC 2010 INIT B: _DISTRO is 001
==> Thu Aug 26 16:10:40 UTC 2010 INIT B: RAW _DOMAIN is host.aegir.cc
==> Thu Aug 26 16:10:40 UTC 2010 INIT B: Run standard installer
==> Thu Aug 26 16:10:50 UTC 2010 INIT B: Drush seems to be functionning properly
==> Thu Aug 26 16:10:50 UTC 2010 INIT B: Installing provision backend in /data/disk/host/.drush
==> Thu Aug 26 16:10:51 UTC 2010 INIT B: Build Aegir base
==> Thu Aug 26 16:11:00 UTC 2010 INIT STATUS: INIT
==> Thu Aug 26 16:11:09 UTC 2010 INIT B: Deploying hostmaster application init
==> Thu Aug 26 16:11:32 UTC 2010 INIT B: Deploying hostmaster application done
==> Thu Aug 26 16:11:32 UTC 2010 INIT B: Simple check if Aegir install is successful
==> Thu Aug 26 16:11:41 UTC 2010 INIT INFO: Aegir setup test OK
==> Thu Aug 26 16:11:41 UTC 2010 INIT B: Enhance Aegir UI
==> Thu Aug 26 16:11:57 UTC 2010 INIT B: Aegir global.inc file
==> Thu Aug 26 16:11:57 UTC 2010 INIT A: Run standard post-installer
==> Thu Aug 26 16:12:06 UTC 2010 INIT A: Run standard post-installer completed
==> Thu Aug 26 16:12:06 UTC 2010 INIT A: Set permissions for all
==> Thu Aug 26 16:12:15 UTC 2010 INIT A: Switch user and run Distros build

Do you want to install also some ready to use platforms? [Y/n] Y
==> NOTE: Please choose the platforms/distributions you want to have included

==> Thu Aug 26 16:12:24 UTC 2010 CLONE C: Aegir automated install script part B
==> Thu Aug 26 16:12:24 UTC 2010 CLONE C: Create directories
 
Drupal 5 Pressflow - http://pressflow.org [Y/n] n
==> STATUS: Drupal 5 Pressflow installation skipped
 
Acquia - http://bit.ly/acquiadrupal [Y/n] Y
==> STATUS: Acquia install in progress, please wait...
==> STATUS: Acquia installation completed
 
Open Atrium - http://openatrium.com [Y/n] Y
==> STATUS: Open Atrium install in progress, please wait...
==> STATUS: Open Atrium installation completed
 
Open Atrium dev - http://openatrium.com [Y/n] Y
==> STATUS: Open Atrium live head version install in progress, please wait...
==> STATUS: Open Atrium live head installation completed
 
Prosepoint - http://prosepoint.org [Y/n] n
==> STATUS: Prosepoint installation skipped
 
Drupal 6 Pressflow - http://pressflow.org [Y/n] Y
==> STATUS: Drupal 6 Pressflow install in progress, please wait...
==> STATUS: Drupal 6 Pressflow installation completed
 
Ubercart - http://ubercart.org [Y/n] Y
==> STATUS: Ubercart install in progress, please wait...
==> STATUS: Ubercart installation completed
 
Open Publish - http://openpublishapp.com [Y/n] Y
==> STATUS: Open Publish install in progress, please wait...
==> STATUS: Open Publish installation completed
 
Managing News - http://managingnews.com [Y/n] Y
==> STATUS: Managing News install in progress, please wait...
==> STATUS: Managing News installation completed
 
Managing News dev - http://managingnews.com [Y/n] Y
==> STATUS: Managing News live head version install in progress, please wait...
==> STATUS: Managing News live head version installation completed
 
Feature Server - http://bit.ly/fservermore [Y/n] Y
==> STATUS: Feature Server install in progress, please wait...
patching file singular/style.css
==> STATUS: Feature Server installation completed
 
Drupal Commons - http://acquia.com/drupalcommons [Y/n] Y
==> STATUS: Drupal Commons install in progress, please wait...
==> STATUS: Drupal Commons installation completed
 
Drupal 7 - http://drupal.org/project/drupal [Y/n] Y
==> STATUS: Drupal 7 install in progress, please wait...
==> STATUS: Drupal 7 installation completed
==> Thu Aug 26 16:17:04 UTC 2010 CLONE C: remove some default themes for all
==> Thu Aug 26 16:17:13 UTC 2010 CLONE C: Save & Verify Platforms
==> Thu Aug 26 16:17:25 UTC 2010 CLONE C: All done!
==> Thu Aug 26 16:17:25 UTC 2010 INIT A: Clean up old and install files
==> Thu Aug 26 16:17:34 UTC 2010 INIT A: Add cron task
==> Thu Aug 26 16:17:43 UTC 2010 INIT A: Touch install logs
==> Thu Aug 26 16:17:43 UTC 2010 INIT A: Start the cron now
==> Thu Aug 26 16:17:44 UTC 2010 INIT STATUS: All done!
==> Thu Aug 26 16:17:54 UTC 2010 INIT BYE!
v246a:~#

Comments

First obvious Q & A

Posted by omega8cc on August 26, 2010 at 5:24pm

Q: How to upgrade server OS, MariaDB, PHP, Nginx and the main Aegir instance?
A: Run again as root $ bash BARRACUDA.sh.txt.

Q: How to add more instances of Aegir?
A: Edit domain and username and run as root $ bash OCTOPUS.sh.txt.

Q: How to upgrade the Aegir instance(s) and add new releases of platforms?
A: Use the same domain and username of the instance and run again as root $ bash OCTOPUS.sh.txt.

How can we make it easier?

Same instance and user in BARRACUDA.sh.txt and OCTOPUS.sh.txt

Posted by Chipie on September 1, 2010 at 4:07pm

Can I use the same Aegir instance (user:aegir., domain aegir.example.com) in BARRACUDA.sh.txt and OCTOPUS.sh.txt? Or do you recommend to set up a separate user and instance for OCTOPUS.sh.txt.?

You can't. It is designed to

Posted by omega8cc on September 2, 2010 at 12:14am

You can't. It is designed to be separate instances, while that one created by Barracuda is called core Hostmaster due to the role in the Nginx configuration it plays here. Barracuda installer is designed to install and upgrade your OS environment and core HM, and working instances should be created using Octopus, which is designed to install and upgrade only non-core Hostmasters and hosted there platforms with shared code.

If it will create another

Posted by anantagati on September 11, 2010 at 4:20pm

If it will create another Aegir instance, what should I use Aegir instance created by Barracuda for? And for what should I use Aegir created by Octopus?

If I want to have just few sites with few platforms, isn't for me better to just have one Aegir (made by Barracuda) for it?

The core Hostmaster holds the

Posted by omega8cc on September 14, 2010 at 3:33pm

The core Hostmaster holds the main Nginx configuration files, while Octopus instances are using includes to run its web sites using main Barracuda configuration.

The reason we designed it to be separate is to guarantee compatibility for upgrades with all old installs, be default using /var/aegir directory and "aegir" as user, probably with platforms in many different directories.

This is why Hostmaster instance created by Barracuda comes without any platforms.

The Octopus is designed to allow many non-core Hostmaster instances on the same server with shared code base, for easier debugging, hosting of separated Aegir environments, so you no longer have to worry about not breaking your the only Aegir instance.

It is recommended to leave as is (not use) the core Hostmaster (unless you have some old platforms to be imported there) and use only Octopus instances to deploy and manage sites.

If anything will go wrong, you can always create another Octopus instance and import there your live platforms and sites.

Thanks for the answer. It is

Posted by anantagati on September 14, 2010 at 8:30pm

Thanks for the answer. It is now clear for me.

If I will be more Octopus instances, what is best way if I will want to uninstall one?

To uninstall any Octopus

Posted by omega8cc on September 17, 2010 at 1:46pm

To uninstall any Octopus instance: disable and delete all sites hosted there to delete not used databases and files, backup all files from /data/disk/octopususer just in case and move this directory away, remove this instance symlink from /var/aegir/config/server_master/nginx/platform.d, restart Nginx, and it is done.

This isn't quite clear

Posted by tseven on January 27, 2011 at 7:40pm

To uninstall any Octopus instance: disable and delete all sites hosted there to delete not used databases and files

It reads as if deleting the files from the paths mentioned and restarting nginx will also remove the unused databases. From my experience, this is not so. Is there a maintenance script which will automatically run, removing the databases? Or perhaps they will be removed the next time I run the Octopus shell script.

Clarification on this would be great.

Thank you.

EDIT:
After deleting those files, you still have two system users: username and username.ftp as well as the site's database hanging around.
Are there any plans on creating a script/drush command to completely remove octopus instances? Even better would be an option to archive/backup and delete which would create a tarball which could be used later to restore the client.

With some of the latest

Posted by rpsu on July 9, 2014 at 8:44am

With some of the latest updates related to multiple php-versions you also might need to remove php-configurations which point to removed /data/disk/[octopususer] -directories. If not, php-fpm will fail to start - and neither of barracuda or octopus script seems to notice that.

XX is your php-version:
rm /opt/phpXX/etc/pool.d/[octopususer].conf

--
Perttu Ehn

Changelog on the AEgir day in CPH

Posted by omega8cc on August 27, 2010 at 3:22am

### Fri, 2010-08-27 04:15 - Edition 0.4-alpha12-A12.B

Added/Fixed: (upgrade optional)

1. Octopus now allows to install or upgrade only Aegir
   hostmaster instance without any platforms added.

2. Enabled again early exit on the first error to avoid
   confusing cascade of errors if something went wrong.

3. Both Barracuda and Octopus runs now faster.


### Thu, 2010-08-26 19:30 - Edition 0.4-alpha12-A12.A

Added/Fixed: (upgrade from previous versions recommended)

1. Barracuda now includes multicore Apache Solr Search,
   Redis and Memcache.

2. Barracuda now can upgrade packages selectively.
   Just run it again to upgrade the system and the main
   Aegir instance.
   
3. Octopus can create many sub-instances on the same
   server, each with different set of platforms, but
   with ability to share the code between instances,
   so you can use this system even on the low end VPS.
   
4. Chive database manager added by default with db.
   subdomain (may require dns entry or wildcard).


### Thu, 2010-08-26 08:55 - Edition 0.4-alpha12-A12.A

Added/Fixed: (upgrade from previous versions recommended)

1. By defult Aegir 0.4-alpha12 with Drush 3.3
   is now installed.

2. Introduced new Octopus and Barracuda installers.
   See README.txt for more information.
   Both are in pre-alpha debugging phase.
   
3. All installers code and helpers now hosted on GitHub.

Fantastic omega8cc.

Posted by stevenwood on August 26, 2010 at 7:56pm

Fantastic omega8cc.

sweet mercy

Posted by flailingmaster on August 27, 2010 at 12:27am

this looks amazing!

Great work omega8cc

Posted by jaymiejones86 on August 27, 2010 at 11:35pm

This looks really great. Top work so far on this. I think this will be integral for further adoption, even though installing Aegir isn't the hardest task this makes a lot of the initial setup and ongoing maintenance much easier.

But for some reason, on my MediaTemple VE server (running Ubuntu 10.04) I get a DNS test ok but it prints out this "Your hostname appears to be - are you sure it's a valid FQDN hostname?". and also does not find my IP address.

Now thats all fine if I manually add the IP, hostname, etc, but I get a bunch of errors regarding perl, EG

-Can't exec "/tmp/libssl0.9.8.config.141551": Permission denied at /usr/share/perl/5.10/IPC/Open3.pm line 168.
open2: exec of /tmp/libssl0.9.8.config.141551 configure 0.9.8g-16ubuntu3.1 failed at /usr/share/perl5/Debconf/ConfModule.pm line 59

and so on and so fourth with those kind of perl errors with an exit status of 255. All of this is being run via root btw, so there shouldn't be any permission errors.

And the final output once it completely fails:

Displaying the last 15 lines of /var/tmp/aegir-install.log to help troubleshoot this problem:
  error adding signet.pl/signet_pca2_pem.crt
  error adding signet.pl/signet_pca3_pem.crt
  error adding signet.pl/signet_rootca_pem.crt
  error adding signet.pl/signet_tsa1_pem.crt
  error adding spi-inc.org/spi-ca-2003.crt
  error adding spi-inc.org/spi-cacert-2008.crt
  error adding telesec.de/deutsche-telekom-root-ca-2.crt
failed.
dpkg: error processing ca-certificates-java (--configure):
 subprocess installed post-installation script returned error exit status 1
Processing triggers for libc-bin ...
ldconfig deferred processing now taking place
Processing triggers for python-support ...
Errors were encountered while processing:
 ca-certificates-java

==> Fatal Error Occurred: Installation failed: 0
==> Cannot continue installation.

Does anyone have any ideas? I may be missing something quite simple. Sorry if this is in the wrong section for issues.

Thanks in advance people and keep up the awesome work.

Can't exec /tmp/...

Posted by davej on August 28, 2010 at 11:09am

@jaymiejones86: Could it be that /tmp is mounted with the "noexec" option?

Dave

noexec option

Posted by jaymiejones86 on August 28, 2010 at 11:36am

Hi davej,

I am not 100% I'm still fairly new to server management, but I looked in fstab but cant see any antry for /tmp so I cant see a noexec option there...

noexec option

Posted by davej on August 28, 2010 at 11:49am

These commands might shed some light:

mount
df
df /tmp

Dave

Good catch. We will stop

Posted by omega8cc on August 28, 2010 at 4:27pm

Good catch. We will stop using /tmp to avoid this problem when it is noexec mounted. Some hosts has this security feature enabled. I have seen this in ServInt used by default.

Barracuda/Octopus make me

Posted by entrigan on August 28, 2010 at 12:54am

Barracuda/Octopus make me smile. For those considering this, it runs noticeably faster than a vanilla lamp/aegir install. (at least on a linode 512 environment)

How I installed it (was not entirely obvious to me at first):

Get a clean Ubuntu install
wget BARRACUDA.sh.txt and OCTOPUS.sh.txt (here).
Edit email in BARRACUDA.sh.txt
run "bash BARRACUDA.sh.txt"
wait a while
You now have a working web server and aegir install (see the email BARRACUDA sent you).

To get OCTOPUS with preloaded platforms:

Edit email and hostname of OCTOPUS.sh.txt
run "bash OCTOPUS.sh.txt"
Follow prompts

Thanks! We need to improve

Posted by omega8cc on August 28, 2010 at 4:29pm

Thanks! We need to improve the docs and how-to. I hope there will be soon also video posted to make it even easier.

I Would like to help with

Posted by mindtrades on October 10, 2010 at 12:16am

I Would like to help with documentation parts and videos if you wish.

Vin

Posted by omega8cc on October 16, 2010 at 1:51pm

Any help is appreciated. We will have a community site very soon and there will be also a screencasts/video section to share experience, how-to's and learn from each other. If you have some ideas etc, feel free to contact me via contact form here. Thank you!

Can help with How to's and videos

Posted by mindtrades on October 18, 2010 at 3:18pm

Hi Grace
I can help with documentation,how-to and videos and become involved with your great project.
Do let me know what you want to cover first and i would start with it

errors on octopus install

Posted by Ari Gold on September 3, 2010 at 9:18pm

i have installed barracuda as aegir.mydomain.com and installing octopus on host.mydomain.com with username "host" and get errors:

==> STATUS: This script is ran as a root user.
NEW AEGIR setup in progress..
_DISTRO is 001
RAW _DOMAIN is host.mydomain.com

==> START -> checkpoint:

Your e-mail address appears to be [...] - is that correct?
Your Aegir control panel for this instance will be available at http://host.mydomain.com.
Your Aegir user for this instance will be host.

Do you want to proceed with the setup? [Y/n]
==> Fri Sep 3 20:55:31 UTC 2010 INIT A: Welcome Linux server.mydomain.com [...] #4 SMP Wed Aug 11 15:44:25 UTC 2010 x86_64 GNU/Linux
==> Fri Sep 3 20:55:35 UTC 2010 INIT A: Aegir automated install script part A
==> Fri Sep 3 20:55:35 UTC 2010 INIT A: Stop the cron and wait 3 seconds..
==> Fri Sep 3 20:55:40 UTC 2010 INIT A: Checking status..
==> Fri Sep 3 20:55:44 UTC 2010 INIT A: _STATUS is INIT
==> Fri Sep 3 20:55:44 UTC 2010 INIT A: Create directories and/or set good permissions
==> Fri Sep 3 20:55:48 UTC 2010 INIT A: Add user
==> Fri Sep 3 20:55:53 UTC 2010 INIT A: Prepare scripts
==> Fri Sep 3 20:55:57 UTC 2010 INIT A: pre-install start
==> Fri Sep 3 20:56:05 UTC 2010 INIT A: pre-install completed
==> Fri Sep 3 20:56:09 UTC 2010 INIT A: Switch user and run AegirSetupB
==> Fri Sep 3 20:56:13 UTC 2010 INIT B: Aegir automated install script part B
==> Fri Sep 3 20:56:13 UTC 2010 INIT B: Checking status..
==> Fri Sep 3 20:56:17 UTC 2010 INIT B: Create dirs and/or set good permissions
==> Fri Sep 3 20:56:21 UTC 2010 INIT B: NEW AEGIR setup in progress..
==> Fri Sep 3 20:56:21 UTC 2010 INIT B: _DISTRO is 001
==> Fri Sep 3 20:56:21 UTC 2010 INIT B: RAW _DOMAIN is host.mydomain.com
==> Fri Sep 3 20:56:21 UTC 2010 INIT B: Run standard installer
==> Fri Sep 3 20:56:26 UTC 2010 INIT B: Drush seems to be functioning properly
==> Fri Sep 3 20:56:26 UTC 2010 INIT B: Installing provision backend in /data/disk/host/.drush
==> Fri Sep 3 20:56:28 UTC 2010 INIT B: Build Aegir base
==> Fri Sep 3 20:56:32 UTC 2010 INIT B: INIT
==> Fri Sep 3 20:56:36 UTC 2010 INIT B: Deploying hostmaster application init
==> Fri Sep 3 20:56:38 UTC 2010 INIT A: Aegir Hostmaster instance install completed
==> Fri Sep 3 20:56:38 UTC 2010 INIT A: post-install start
mv: cannot stat nginx_simple_include.conf': No such file or directory mv: cannot statnginx_advanced_include.conf': No such file or directory
chgrp: cannot access /data/disk/host/aegir/distro/001/sites/host.mydomain.com/files': No such file or directory chgrp: cannot access/data/disk/host/aegir/distro/001/sites/host.mydomain.com/settings.php': No such file or directory
/opt/tmp/nginx-for-drupal/aegir/barracuda/AegirSetupA.sh.txt: line 300: cd: /data/disk/host/aegir/distro/001: No such file or directory
==> Fri Sep 3 20:56:42 UTC 2010 INIT A: post-install completed

......
then it installs platforms without any errors. I can login to aegir.mydomain.com but host.mydomain.com just gives a "Welcome to nginx!" message

i did everything on a clean lucid install logged in as root. should i have added a linux user "host"?

changed user from host to admin and worked!

Posted by Ari Gold on September 3, 2010 at 10:06pm

I changed user from "host" to "admin" and didn't get any of the previous errors.

Same Octopus install errors, previous fix doens't work

Posted by Ari Gold on September 17, 2010 at 9:09pm

So i'm trying a clean install, using the same process as before, and I'm running into the same problem where Barracuda installs properly to aegir.mydomain.com but Octopus fails with same errors like these:

AegirSetupC.sh.txt: line 818: cd: /data/disk/admin/aegir/distro/001/sites/garden.mydomain.com: No such file or directory

I tried host.mydomain.com for Octopus and now garden.mydomain.com and neither works. Since the thread is here I thought to continue it, but I'm thinking about posting an issue on git. The previous fix (changing Octopus user to admin didn't work this time)

/data/disk/admin/aegir is empty....

What might I be doing wrong?

If only that was in my case

Posted by jaymiejones86 on August 28, 2010 at 2:58am

If only that was the case for me entrigan. Maybe it something to do with MediaTemple but it doesnt. Very curious. Might try it on a local VM. But if anyone has any ideas in regards to my issue, it would be greatly appreciated.

Having tried to install both

Posted by amstercad on August 30, 2010 at 6:34pm

Having tried to install both beta11 and beta12 on a VirtualBox virtual machine, I cannot report any success or joy either.

Nevermind the process has also been reliably easy for me on a hosted VPS (thanks Scripters!!), I cannot repeat the install successfully using the same base Ubuntu 10.4 server, with only OpenSSH and networking installed. Using either local VirtualBox or a remote ssh client.

Maybe it is only a VirtualBox thing I am seeing, but VMware also seems to fail: http://groups.drupal.org/node/84074#comment-270274

Maybe it is because the address is non-route-able 192.168.x.x and behind a firewall with no open ports(?); although I did configure the local DHCP router with the server address, along with the public DNS as well as the hosts file. Still I consistently receive these errors:

==> Mon Aug 30 20:20:49 CEST 2010 UPGRADE A: Prepare scripts
==> Mon Aug 30 20:20:53 CEST 2010 UPGRADE A: Run standard pre-installer
==> This server does not have a hostname that resolves to an IP address

hostname and hostname -f both resolve the full host.domain perfectly btw. My hosts file looks like:

127.0.0.1 localhost
192.168.15.75 ubuntu-aegir ubuntu-aegir.example.com
123.456.789.123 ubuntu-aegir ubuntu-aegir.example.com

It doesn't seem to matter whether VirtualBox Additions are installed or not. At least VM snapshots allow for easy, reliable, often reptitive testing. Both Barracuda and Octopus scripts fail, and I entered the IP address (192.168.x.x.) and FQDN in Barracuda manually, tried it without, etc.

Anyone have other suggestions to try?

Barracuda in the local

Posted by omega8cc on August 30, 2010 at 8:21pm

Barracuda in the local environment is not fully supported. It is considered as an experimental use case only. This is because Aegir 0.4-alpha9 introduced new requirements for FQDN of the hostname, database server listening on public IP and not on localhost only - everything due to introduced remote server/sites management features.

Some people reported (in the GitHub issue queue) success with Barracuda on localhost with proper DNS setup on the host machine, but we can't support it since there are too many ways to do the local DNS wrong, so every issue related to running Barracuda in the local VM is considered as a support request and not as a bug submission.

If someone has success with running it locally, please post how-to and patches if required for review. Your support is highly appreciated.

Barracuda works, Octopus doesn't

Posted by aireworth on August 30, 2010 at 10:53pm

I have installed Aegir using Barracuda in a locally running vmware virtual machine by installing dnsmasq. Dnsmasq is a pseudo-ish DNS sever designed to resolve NATted machines behind firewalls on local networks, and it provides proper DNS resolution for local machine. It can be used to solely for the machine it is installed on (although it assumes its being installed on a firewall/router), and thus, installed on a virtual machine, allows enough DNS resolution to allow Barracuda to run.

apt-get install dnsmasq

then edit /etc/dnsmasq.conf - its well commented, but I altered or uncommented:-

 
bogus-priv
resolv-file=/home/colin/nameservers
strict-order
local=/myfakedomain/
expand-hosts
domain=myfakedomain

You then need to have the only entry in /etc/resolv.conf as nameserver 127.0.0.1
and create the file /home/colin/nameservers (or whatever matches in dnsmasq.conf). In this you should put the name of you external nameservers e.g
nameserver 208.67.222.222 (This is OpenDNS's server - it didn't work when I used the nameserver given by VMware).
In /etc/hosts put ip address of the virtual machine and localhost - eg

 127.0.0.1 localhost.localdomain  localhost
192.168.xx.xxx   ubuntu.myfakedomain  ubuntu

(replace the xxxs)

Restart networking, restart dnsmasq. If this is working then the command host 192.168.xx.xxx (again replacing the xxxs) should return ubuntu.myfakedomain. Then run Barracuda, pass the DNS test and install!

Aegir installs, and I can get to the site. All looks well, but I can't install new sites as there is "no valid platform". I'm new to aegir, I suppose I'll have to find out how it works.

One gotcha was that if using DCHP (as is the default in my virtaul machines) resolv.conf gets rewritten on EVERY reboot - thus loosing the setting 127.0.0.1 that's been placed there. There is a fix for this by editing some dchp config file somewhere but so far I've just re-edited it by hand.

So Barracuda completes without error, but when I then run Octopus, I get an error from mysqladmin with access denied for root at localhost with password NO, then some missing directories /data/disk/host/aegir/distro/001 - no such file or directory.
Any ideas?

Many thanks

Colin

Thanks, this helped! ...How to create new Platforms

Posted by amstercad on August 31, 2010 at 6:55am

Thanks Colin, your information helped me diagnose the hosts file, which I changed to match yours; it worked! Barracuda tries to install now, but Octopus still fails exactly as before. BTW, I use DD-WRT routers religiously, which make use of DNSmasq also. DD-WRT has a very good wiki full of the common configuration recipes like DNSmasq.

Actually SMTP relay settings are delaying the install now, but I'll sort that out today. Thanks!

Installing platforms to use for sites is really easy. Let's say you wanted both a production (RED) and development (BLUE) platform to use for OpenAtrium, and /var/aegir/platforms holds everything. Don't do this as root, but as your aegir user.

 #Switch to aegir user
sudo su -s /bin/bash - aegiruser
cd /var/aegir/platforms

#Download and uncompress OpenAtrium
wget http://openatrium.com/sites/openatrium.com/files/atrium_releases/atrium-...
tar xvfz atrium-1-0-beta8.tgz

# Rename the uncompressed folder, using the MV (move) command to something useful:
mv atrium-1-0-beta8 atrium-1-0-beta8-RED

# Make a copy
cp -r atrium-1-0-beta8-RED atrium-1-0-beta8-BLUE

Switch to your Drupal/Aegir web browser interface, and Create Platform
Give your platform a name like OpenAtrium Production RED
Now copy and paste the path /var/aegir/platforms/atrium-1-0-beta8-RED and allow a little time for Aegir to verify the platform. Afterwards if all goes well, your platfform is now available to create, clone, or migrate sites.

TIP: It is easy to 'touch' files/folders as the root user, which causes permissions errors, which isn't always obvious in operating Drupal. So be aware of this, and if/when you need to, remember to

chown -R aegiruser:aegiruser /var/aegir/platforms/folder/in/question

Thanks for the info, but

Posted by aireworth on September 2, 2010 at 12:22am

Thanks for the info, but maybe I wouldn't need it now as I have Octopus working (see below). The problem is now one of too much choice rather than too little.

The hosts file format comes from trying to install Zimbra on a machine not directly connected to the internet. Largely unsuccessful, and now superseded, thankfully. But they were very keen to have the host file in that format, with only those entries, fqdn first, short name second, localhost as first entry. Seems to work well in many situations.

Did you get your relay host sorted? I relay through 1and1.co.uk where my mail is hosted. Their setup requires authentication and I had to setup 4 things:-

a) the name of the relayhost
b) a hashed username/password file for authentication - using my actual 1and1.co.uk mail username and password
c)smtp_sasl enabled for authentication
d)myorigin set to a real domain (presumably the same domain as your authentication username uses will always work)
So in /etc/postfix/main.cf we have :-

 relayhost = [auth.smtp.1and1.co.uk] 
smtp_sasl_auth_enable = yes
smtp_sasl_password_maps = hash:/etc/postfix/sasl_passwd
smtp_sasl_security_options = 
myorigin = /etc/mailname

In /etc/mailname a single line with a real resolving MX domain 

In /etc/postfix/sasl_passwd
auth.smtp.1and1.co.uk  username@mymaildomain:mypassword

Hash your sasl_passwd file

postmap /etc/postfix/sasl_passwd

Refresh postfix
postfix reload

Job done. Just remember any mail already queued to go will not get through, (for me at any rate) as it will be rejected as having an invalid FROM address - that's why you need to set myorigin. The queue can be flushed with postsuper -d ALL

Colin

That part about SMTP relay in

Posted by omega8cc on September 2, 2010 at 2:33am

That part about SMTP relay in Barracuda refers only to situation when your host requires using their own SMTP gateway and it doesn't require any authentication enabled, since they know you are sending it from their network. There are no plans to add auth version, since most of the hosts don't block outgoing connections for SMTP and support for local installs is not a priority, rather nice to have feature. Thanks for sharing your how-to, it can be useful for those trying local install and having Internet access with blocked outgoing SMTP.

I'm probably not your typical

Posted by aireworth on September 2, 2010 at 7:09pm

I'm probably not your typical user case. I don't have access to a whole spare VPS for testing, Your scripts look like an excellent way of dipping my toe in the Aegir water. So for me it has to be installed in a virtual machine on a local network. I don't have a static IP or MX record at home, so it has to be relayed, and obviously the relayer is going to need more verification under these circumstances.

Did you get your relay host

Posted by amstercad on September 3, 2010 at 10:47am

Did you get your relay host sorted? I relay through 1and1.co.uk where my mail is hosted.

No, although I tried very hard and checked and re-checked and triple-checked everything many times over. One thing I tried was Google Apps for Domains, and gmail servers as a relay host. (Dreamhost.com smtp also) This google MX server should work, along with the username/password, as you have documented: aspmx.l.google.com

I wrote that it should work, however I am not seeing the error in my setup.

I agree it is fairly normal to want to develop & test using low-cost Virtual Machines, possibly locally on a LAN. I don't want to risk breaking my production server with untested stuff, and can't really see the point of buying another VPS instance just for testing & development. For example, I would like to get awstats running reliably, somehow, before I try to install it on any production NGINX VPS. This is why I've been working so hard to get Aegir/Barracuda running on a Virtual machine. This is also why I try to invest in a quality LAN configuration.

I'm not sure what I'll try next; I'm evaluating my options. So far, I cannot pass the SMTP check. But it is encouraging to see others have had success with virtual machine installations.

I think we will make this

Posted by omega8cc on September 3, 2010 at 11:46am

I think we will make this SMTP test optional, to not block local installs due to not working outgoing port 25 connections.

This is very welcome news,

Posted by amstercad on September 3, 2010 at 1:27pm

This is very welcome news, thank you very much! Drupal itself processes outgoing mail very well, so I hope something similar is possible.

SMTP check is now optional:

Posted by omega8cc on September 3, 2010 at 1:59pm

SMTP check is now optional: http://github.com/omega8cc/nginx-for-drupal/commit/73e4fec75042e79d7bdbf...

Virtual Machine Installation Joy

Posted by amstercad on September 8, 2010 at 11:46pm

I can confirm that with the SMTP check off, Barracuda Aegir/NGINX installs.

But not until I created a DNS A-record for chive at db.example.com, along with the a-record for Aegir web panel itself, etc. One must do this.

In my case, it is possible my ISP is blocking port 25, I'm not sure. I do know I have two SMTP possibilities using port 587 (and also port 465). It would be nice if the port was configurable, and not locked to 25. I tried various incantations to specify port 587, but failed.

I can confirm access to port 587 smtp (but not 25) with a simple command line test as follows:

telnet smtp.gmail.com 587

This is how I configured DNSmasq using DD-WRT, to get a single virtual host to answer to multiple domain names:

From the DD-WRT web interface, click the Services main-tab, and then the Services sub-tab, (never-mind the redundancy).

'LAN Domain' is used to assign a private, internal domain name to the fire-walled LAN, like example.com. (If you don't configure your public WAN to use your fire-walled servers, this remains private, while being functional)

Normally, you use MAC addresses to assign Host names to IP addresses for machines that have Mac addresses. Nothing new here.

In the case of serving websites from the single Aegir managed server, use 'Additional DNSMasq Options'.

Enter the Mac address of the virtual machine and matching IP address manually as follows, along with all the websites that share it, like this example:

dhcp-host=03:BC:21:6C:23:8B,192.168.2.99,24h
address=/websiteA.example.com/192.168.2.99
ptr-record=websiteA,192.168.2.99
address=/websiteB.example.com/192.168.2.99
ptr-record=websiteB,192.168.2.99

Optionally, you can decide to forward DD-WRT firewall ports 80, 443, etc., directed to the Aegir server on the LAN, of course.

Thank you very much NGINX scripters, along with the Aegir devs!

But if you are failing at the

Posted by aireworth on September 3, 2010 at 2:03pm

But if you are failing at the smtp check, as opposed to failing to correctly relay mail (as was my case) then surely its a port 25 problem on the virtual machine to the host, or the host to the internet, or on the way back, or perhaps a failure to reslove smtp.host8.biz correctly, is it not? (I might not fully understand how the SMTP check works but it looks like its testing whether your machine can connect to a remote mail server and get a response back).

Can you ping from your virtual machine to smtp.host8.biz ? Are all ports 25 open on the way out and the way back in?

Colin

Thanks

Posted by Chipie on August 31, 2010 at 4:49pm

Thanks for your help with dnsmasq. I was able to run the Barracuda and Octopus installer without any errors.

Ooops- Don't restart networking

Posted by aireworth on September 1, 2010 at 10:30pm

I just discovered an error in my description of what worked above. Resolv.conf gets overwritten every time networking gets restarted. So don't restart networking, (and no need to restart dnsmasq either)
EDIT
Yes you do need to restart dnsmasq - otherwise the changes in /home/colin/nameservers don't take.

And you can prevent the changes to /etc/reslov.conf on a reboot or networking restart by using chattr to make it read only -

chattr +i /etc/resolv.conf

Crude, but (as is often the case) effective.

Colin

Prepending resolv.conf w/ local IP to avoid overwrite issue

Posted by dmartens on September 4, 2010 at 6:15pm

@Colin - thanks for the heads up on using dnsmasq for local dns. That was an excellent starting point for experimentation on my part.

Your notes above acknowledged the fact that resolv.conf gets overwritten each time networking gets restarted. For those that need a dynamic IP, like me, this presents a problem especially since I must restart networking pretty regularly (as I migrate from one coffeeshop to another). My use case involves running Aegir from a virtualmachine on my laptop, but interfacing to a server on the Internet.

You mention precluding overwriting of the resolv.conf file and you also allude to the existence of a DHCP option for achieving the same result.

Regarding the second scenario: I figured out what that DHCP option was, and verified that it worked.

Steps:
1. edit the configuration file for the DHCP client to prepend the IP address of our local nameserver (e.g., 127.0.0.1) to the resolv.conf file each time networking gets restarted. The filename of this configuration file is /etc/dhcp3/dhclient.conf.
2. to this file, add the following line: "prepend domain-name-servers 127.0.0.1;" (you'll probably find that this line already exists in dhclient.conf, but that the line commented out).
3. save the file and restart networking ("sudo /etc/init.d/networking restart")

Thereafter, a quick check of resolv.conf confirms that 127.0.0.1 is placed first in the resolv.conf file (exactly what we want for local name resolution) and that the DNS nameservers acquired from the DHCP server at the coffeeshop or wherever follow 127.0.0.1 (to ensure that global name resolution still works). Also, "dig (localhostname)" and "dig google.com" confirm that local and global names get resolved properly.

In the spirit of moving the local name resolution issue forward, I offer this small tidbit.

Thanks again for your earlier posts ...
Dave

Is it possible for you to

Posted by vedat.kamer on August 28, 2010 at 3:48pm

Is it possible for you to give usage senarios of Octopus script? Is it suitable for providing Aegir hosting?

It is suitable for testing

Posted by omega8cc on August 28, 2010 at 4:38pm

It is suitable for testing and using as a base to create your own fork on github. It is not for creating ready to use hosting service since there is no ready to use setup of ftp/ssh accounts with all required security settings etc. It is designed to help small drupal shops who want to use Aegir (and not to learn Aegir) to manage their own environment and their clients sites, but without direct access to files for other people. But it is possible we will add more features, like http://drupal.org/project/uc_hosting very soon.

Question

Posted by zkrebs on August 30, 2010 at 10:09pm

I performed the upgrade from B. Aegir HEAD and it worked flawlessly, upgraded everything as stated. Except, I was never offered a chance to install any Drupal distributions/platforms. Is this by design? Clearly I can download them myself, and omega8cc you generously gave them to me in a nice package before. Was just wondering..perhaps for future people? I also have to say this is basically badass, awesome, wonderful, great, supreme....etc. :)

quick question:

How do I connect my site to Apache SOLR that comes with the upgrade? I've done this before...what would the hostname, port, etc. all be?

How do I connect to chive? I have wildcard DNS..but chive.mydomain.com says "Welcome to Nginx"

By default it is set to

Posted by omega8cc on September 1, 2010 at 1:00am

By default it is set to db.mydomain.com. It is displayed on the first checking point on install.

Ah, for Solr readme see docs/SOLR.txt

And for distributions/platforms look at Octopus installer.

I can't load chive either.

Posted by toddgeist on September 1, 2010 at 7:05am

I can't load chive either. Chive appears to be install and the virtual host was added to nginx. And I restarted nginX. But I still get "welcome to nginx" instead of chive whenever I try go there.

any clues?

THanks

Todd

Check twice if you are using

Posted by omega8cc on September 2, 2010 at 12:20am

Check twice if you are using subdomain created by Barracuda. By default it is db.yourdomain.ext. Of course this subdomain should be pointed already to your IP. And we should better document it ;)

turns out it was set to

Posted by toddgeist on September 3, 2010 at 10:28pm

turns out it was set to db...com

so my aegir install was under aegir..com

chive was stuck into

db.aegir..com.

I was looking for db..com

Thanks!!!!

Octopus Problem

Posted by toddgeist on August 31, 2010 at 12:59pm

Hello,

I ran both installers. Barracuda went fine. But when Octopus seems to have failed. There were several file missing, or privilege errors in the log. When I navigate to the new aegir control panel I get nothing but the nginx welcome. and I never got a welcome email. See log below...

My understanding is the Octopus installs another instance of aegir, not into the original aegir instance. Is this true?

If so then the _DOMAIN setting should be something different then the original aegir install that barracuda installed. Is this correct?

And the new USER name should be something other than aegir? Is this true?

Thanks for a great tool!

Here is the log.

=> Tue Aug 31 06:32:28 UTC 2010 INIT A: Welcome Linux aegir.eow-02.com 2.6.35.1-rscloud #4 SMP Wed Aug 11 15:44:25 UTC 2010 x86_64 GNU/Linux
==> Tue Aug 31 06:32:33 UTC 2010 INIT A: Aegir automated install script part A
==> Tue Aug 31 06:32:33 UTC 2010 INIT A: Stop the cron and wait 3 seconds..
==> Tue Aug 31 06:32:38 UTC 2010 INIT A: Checking status..
==> Tue Aug 31 06:32:42 UTC 2010 INIT A: _STATUS is INIT
==> Tue Aug 31 06:32:42 UTC 2010 INIT A: Create directories and/or set good permissions
==> Tue Aug 31 06:32:46 UTC 2010 INIT A: Add user
==> Tue Aug 31 06:32:50 UTC 2010 INIT A: Prepare scripts
==> Tue Aug 31 06:32:54 UTC 2010 INIT A: pre-install start
ERROR 1045 (28000): Access denied for user 'root'@'localhost' (using password: YES)
/usr/bin/mysqladmin: connect to server at 'localhost' failed
error: 'Access denied for user 'root'@'localhost' (using password: YES)'
==> Tue Aug 31 06:33:02 UTC 2010 INIT A: pre-install completed
==> Tue Aug 31 06:33:06 UTC 2010 INIT A: Switch user and run AegirSetupB
==> Tue Aug 31 06:33:10 UTC 2010 INIT B: Aegir automated install script part B
==> Tue Aug 31 06:33:10 UTC 2010 INIT B: Checking status..
==> Tue Aug 31 06:33:14 UTC 2010 INIT B: Create dirs and/or set good permissions
==> Tue Aug 31 06:33:18 UTC 2010 INIT B: NEW AEGIR setup in progress..
==> Tue Aug 31 06:33:18 UTC 2010 INIT B: _DISTRO is 001
==> Tue Aug 31 06:33:18 UTC 2010 INIT B: RAW _DOMAIN is host.eow-02.com
==> Tue Aug 31 06:33:18 UTC 2010 INIT B: Run standard installer
==> Tue Aug 31 06:33:24 UTC 2010 INIT B: Drush seems to be functioning properly
==> Tue Aug 31 06:33:24 UTC 2010 INIT B: Installing provision backend in /data/disk/host/.drush
Initialized empty Git repository in /data/disk/host/.drush/provision/.git/
Initialized empty Git repository in /data/disk/host/.drush/provision_boost/.git/
==> Tue Aug 31 06:33:26 UTC 2010 INIT B: Build Aegir base
==> Tue Aug 31 06:33:30 UTC 2010 INIT B: INIT
==> Tue Aug 31 06:33:34 UTC 2010 INIT B: Deploying hostmaster application init
==> Tue Aug 31 06:33:36 UTC 2010 INIT A: Aegir Hostmaster instance install completed
==> Tue Aug 31 06:33:36 UTC 2010 INIT A: post-install start
mv: cannot stat nginx_simple_include.conf': No such file or directory mv: cannot statnginx_advanced_include.conf': No such file or directory
chgrp: cannot access /data/disk/host/aegir/distro/001/sites/host.eow-02.com/files': No such file or directory chgrp: cannot access/data/disk/host/aegir/distro/001/sites/host.eow-02.com/settings.php': No such file or directory
/opt/tmp/nginx-for-drupal/aegir/barracuda/AegirSetupA.sh.txt: line 297: cd: /data/disk/host/aegir/distro/001: No such file or directory
==> Tue Aug 31 06:33:40 UTC 2010 INIT A: post-install completed
==> Tue Aug 31 06:33:40 UTC 2010 INIT A: Set correct permissions
chmod: cannot access `/data/disk/host/aegir/distro/001': No such file or directory
==> Tue Aug 31 06:33:44 UTC 2010 INIT A: Switch user and run Distros build
==> Tue Aug 31 06:33:48 UTC 2010 INIT C: Aegir automated install script part B
==> Tue Aug 31 06:33:48 UTC 2010 INIT C: Create directories
==> Tue Aug 31 06:33:52 UTC 2010 INIT C: Preparing Pressflow core, please wait..

Acquia 6.19-svn P.001 - http://bit.ly/acquiadrupal [Y/n] Y
==> STATUS: Acquia 6.19-svn P.001 install in progress, please wait...
==> STATUS: Acquia 6.19-svn P.001 installation completed

Atrium 1.0-beta8 P.001 - http://openatrium.com [Y/n] Y
==> STATUS: Atrium 1.0-beta8 P.001 install in progress, please wait...
==> STATUS: Atrium 1.0-beta8 P.001 installation completed

Atrium 1.0-beta9-dev P.001 - http://openatrium.com [Y/n] n
==> STATUS: Atrium 1.0-beta9-dev P.001 installation skipped

Drupal 5.23.50 P.001 - http://pressflow.org [Y/n] Y
==> STATUS: Drupal 5.23.50 P.001 install in progress, please wait...
==> STATUS: Drupal 5.23.50 P.001 installation completed

Drupal 6.19 P.001 - http://pressflow.org [Y/n] Y
==> STATUS: Drupal 6.19 P.001 install in progress, please wait...
==> STATUS: Drupal 6.19 P.001 installation completed

Drupal 7.0-alpha6 P.001 - http://drupal.org/project/drupal [Y/n] Y
==> STATUS: Drupal 7.0-alpha6 P.001 install in progress, please wait...
==> STATUS: Drupal 7.0-alpha6 P.001 installation completed

Drupal Commons 1.0 P.001 - http://acquia.com/drupalcommons [Y/n] Y
==> STATUS: Drupal Commons 1.0 P.001 install in progress, please wait...
==> STATUS: Drupal Commons 1.0 P.001 installation completed

Feature Server P.001 - http://bit.ly/fservermore [Y/n] Y
==> STATUS: Feature Server P.001 install in progress, please wait...
==> STATUS: Feature Server P.001 installation completed

MNews 1.0-beta10 P.001 - http://managingnews.com [Y/n] Y
==> STATUS: MNews 1.0-beta10 P.001 install in progress, please wait...
==> STATUS: MNews 1.0-beta10 P.001 installation completed

MNews 1.0-beta11-dev P.001 - http://managingnews.com [Y/n] n
==> STATUS: MNews 1.0-beta11-dev P.001 installation skipped

OpenPublish 2.2.285 P.001 - http://openpublishapp.com [Y/n] Y
==> STATUS: OpenPublish 2.2.285 P.001 install in progress, please wait...
==> STATUS: OpenPublish 2.2.285 P.001 installation completed

Prosepoint 0.35 P.001 - http://prosepoint.org [Y/n] y
==> STATUS: Prosepoint 0.35 P.001 install in progress, please wait...
==> STATUS: Prosepoint 0.35 P.001 installation completed

Ubercart P.001 - http://ubercart.org [Y/n] n
==> STATUS: Ubercart P.001 installation skipped

==> Tue Aug 31 06:38:09 UTC 2010 INIT C: remove some default themes for all
==> Tue Aug 31 06:38:15 UTC 2010 INIT C: Save & Verify Platforms
AegirSetupC.sh.txt: line 701: cd: /data/disk/host/aegir/distro/001/sites/host.eow-02.com: No such file or directory
==> Tue Aug 31 06:38:21 UTC 2010 INIT A: Distributions install completed
==> Tue Aug 31 06:38:21 UTC 2010 INIT A: Clean up old install files
==> Tue Aug 31 06:38:25 UTC 2010 INIT A: Add cron task
==> Tue Aug 31 06:38:29 UTC 2010 INIT A: Touch install logs
==> Tue Aug 31 06:38:29 UTC 2010 INIT A: Start the cron now
==> Tue Aug 31 06:38:30 UTC 2010 INIT A: All done!
==> Tue Aug 31 06:38:35 UTC 2010 BYE!

Was this on a machine with a public IP

Posted by aireworth on August 31, 2010 at 1:16pm

I get exactly the same errors (although I thought mysqlasdmin was denied access with password NO), and exactly the same directories not made, etc when I've tried to install Octopus on a local virtual machine. (See my reply above re DNS settings). Are you getting these errors from a machine with a public IP and DNS settings, or is it local? I thought my errors might be due to me using a virtual machine with a fake domain, but perhaps not.

Regards

Colin

I am using a real domain.

Posted by toddgeist on August 31, 2010 at 3:50pm

I am using a real domain. Although the subdomain does not exist, but there is a wild card entry that should match. Maybe a wildcard won't work? maybe i need to have an A record for the domain.

not sure???

A wildcard subdomain seems

Posted by amstercad on August 31, 2010 at 6:53pm

A wildcard subdomain seems doomed to fail. You probably need solid DNS to succeed with this installation. IMHO if you're setting up complex server environments like Aegir on NGINX with MariaDB, etc., then a $50+ low power/footprint DD-WRT-compatible router, or a recycled PC/server begins to make a lot of sense.

DD-WRT runs on a lot of different brands of routers, and provides a pleasant interface for managing your LAN professionally (which is actually DNSmasq on linux, etc.). Buffalo routers come with DD-WRT pre-installed: http://www.newegg.com/Store/BrandSubCategory.aspx?Brand=1646&SubCategory.... I won't buy a router unless it is listed on the DD-WRT compatibility database.

Nevermind the wireless stuff if you don't need it, but DD-WRT is solid LAN DNS management, and gets my vote. It is the cheapest, and one of of the most important linux servers available to you.

Of course there are lots of other ways too. I'm just saying, you need the right tools for the job.

well I could easily setup the

Posted by toddgeist on August 31, 2010 at 7:34pm

well I could easily setup the subdomain first. Thats not really an issue. I just didn't know I had to do it.

A subdomain and a host (or

Posted by amstercad on August 31, 2010 at 9:28pm

A subdomain and a host (or server) can appear identical in a DNS configuration. Let me know if you need more help.

Same values

Posted by Chipie on August 31, 2010 at 4:46pm

I have used the the values from Barracuda in the Octopus script and had no problems installing all plattforms.

_DOMAIN=aegir.mydomain.com
_USER=admin

Did you set the mysql root password during the Barracuda installer?

i did set the mysql password.

Posted by toddgeist on August 31, 2010 at 7:37pm

i did set the mysql password. Was that not the right thing to do?

So if you use the same host as in barracuda, does that mean that all the platforms are installed into that domain?

Thanks

Did you set it to the

Posted by Chipie on September 1, 2010 at 4:18pm

Did you set it to the suggestion from the BARRACUDA script?

I also set a root password

Posted by tracybarker on September 1, 2010 at 1:54pm

I also set a root password during the BARRACUDA script - is there a place I can enter that password in order to have the OCTOPUS script have access to the database and able to add the platforms?

Properly installed Barracuda

Posted by omega8cc on September 1, 2010 at 2:35pm

Properly installed Barracuda creates two files in your /root directory with root mysql password, so later Octopus doesn't have to read/use password, since it is already there.

After successful installation of Barracuda you should have two important files in your /root directory:

/root/.my.pass.txt should include the root mysql password only.

/root/.my.cnf should include the root mysql password in the correct format for .my.cnf file:

[client]
user=root
password=123mypass

Of course password in both files should be the same.

Sorry for the trouble today,

Posted by omega8cc on September 1, 2010 at 1:03am

Sorry for the trouble today, but it was marked as (test only) so "don't use". There were many live crash tests performed to find weak points in scripts and now it should work as expected for 0.4-HEAD-A12.D Edition, see changelog.

Changelog for Barracuda and Octopus on August 31

Posted by omega8cc on September 1, 2010 at 1:09am

### Tue, 2010-08-31 23:50 - Edition 0.4-HEAD-A12.D

Added/Fixed: (upgrade recommended because it works!)

1. Upgrade of core Hostmaster instance by Barracuda
   and upgrade of Hostmaster instances by Octopus
   finally works as expected.

2. It is now possible to use Barracuda to install
   environment and core Hostmaster instance, to
   upgrade only environment, to upgrade only core
   Hostmaster instance, or both at the same time.
   
3. Octopus now can separately install and/or upgrade
   any non-core Hostmaster instance or any platform
   on any instance, separately, using detailed prompt
   with version numbers and links to distributions
   home pages.

4. New platform Cocomore Drupal added in Octopus:
   http://drupal.cocomore.com  


### Sat, 2010-08-28 20:15 - Edition 0.4-HEAD-A12.C

Added/Fixed: (upgrade recommended)

1. By defult Aegir 0.4-HEAD with Drush 3.3
   is now installed to fix critical issues with
   importing sites. The fix is also available
   as a patch for alpha12:
   http://drupal.org/node/882970#comment-3382542

2. Both Barracuda and Octopus now allow to choose
   if the Aegir Hostmaster will be upgraded or not.

3. Added versions numbers and links to all platforms
   Yes/no prompts.
   
4. /tmp directory no longer used to avoid problems
   due to secure noexec mount.
   
5. Improved readme and docs (in progress).

6. Removed old, no longer supported installer.


### Fri, 2010-08-27 04:15 - Edition 0.4-alpha12-A12.B

Added/Fixed: (upgrade optional)

1. Octopus now allows to install or upgrade only Aegir
   hostmaster instance without any platforms added.

2. Enabled again early exit on the first error to avoid
   confusing cascade of errors if something went wrong.

3. Both Barracuda and Octopus runs now faster.


### Thu, 2010-08-26 19:30 - Edition 0.4-alpha12-A12.A

Added/Fixed: (upgrade from previous versions recommended)

1. Barracuda now includes multicore Apache Solr Search,
   Redis and Memcache.

2. Barracuda now can upgrade packages selectively.
   Just run it again to upgrade the system and the main
   Aegir instance.
   
3. Octopus can create many sub-instances on the same
   server, each with different set of platforms, but
   with ability to share the code between instances,
   so you can use this system even on the low end VPS.
   
4. Chive database manager added by default with db.
   subdomain (may require dns entry or wildcard).


### Thu, 2010-08-26 08:55 - Edition 0.4-alpha12-A12.A

Added/Fixed: (upgrade from previous versions recommended)

1. By defult Aegir 0.4-alpha12 with Drush 3.3
   is now installed.

2. Introduced new Octopus and Barracuda installers.
   See README.txt for more information.
   Both are in pre-alpha debugging phase.
   
3. All installers code and helpers now hosted on GitHub.

Platforms installed as root

Posted by tracybarker on September 1, 2010 at 3:54pm

I've run the OCTOPUS installer and for some reason it's not adding platforms to the aegir installation, and the platforms are stored in /data/all/00x/*, all owned by root instead of aegir. Have I done something wrong? I'm an utter noob, so I wouldn't be surprised.

I ended up copying all of the

Posted by tracybarker on September 1, 2010 at 7:21pm

I ended up copying all of the platforms that were installed in /data/all into /var/aegir/platforms and chown'ing the whole kit and caboodle over to the aegir user, then I manually added the platforms to aegir and it seems to all be working now.

Bad idea. You are trying to

Posted by omega8cc on September 2, 2010 at 12:27am

Bad idea. You are trying to work too hard :)

It is correct. All code is

Posted by omega8cc on September 2, 2010 at 12:26am

It is correct. All code is stored there and owned by root - by design. Your non-core hostmasters with platforms (and with symlinks to the shared code of platforms) live in /data/disk/user/distro/001 and upgrades in 002, 003 etc.

Barracuda works and so does Octopus!

Posted by aireworth on September 1, 2010 at 11:34pm

Barracuda works and so does Octopus!

I was having problems getting Octopus to install properly in a local virtual machine. I was getting 2 errors. Firstly I got the error on one of the aptitude updates of a package being held back. But by pressing enter Barracuda ran, and Aegir installed. Then when I ran Octopus, I got the "mysqladmin access denied error", which resulted in a pile of missing files and directories and a failed install.

So this afternoon I tried again, changing 2 things. Firstly I used a completely bare install of Lucid, with NO apt-get update or apt-get upgrade after installing. (I did install dnsmasq as this was in a local virtual machine). Secondly I ran both scripts, Barracuda and Octopus, from a root shell ( sudo -i) as opposed to running them with sudo bash BARRACUDA.sh.txt

The result was that both errors disappeared, and Octopus installed and is running as intended. Because I changed 2 things at once, its difficult to say which made the difference in the case of either error. I suspect the bare install with no upgrade got rid of the package held back error, and I suspect running in a root shell sorted the mysqladmin access error, but I could be completely wrong.

Anyway, many thanks Grace - you're [code is] lookin' good!

Colin

Today Albert fixed a few old

Posted by omega8cc on September 2, 2010 at 12:33am

Today Albert fixed a few old and new issues, see http://github.com/omega8cc/nginx-for-drupal/commits/master/ and you are right - both installers should be used when you are logged in as root (via su -), no sudo etc tricks.

So root password must be set?

Posted by aireworth on September 2, 2010 at 7:33am

So root password must be set? I used sudo -i which allegedly gives the same result as su - , without the need for a root password.

It doesn't matter how are you

Posted by omega8cc on September 2, 2010 at 9:16am

It doesn't matter how are you get logged in as root to your system. The important part of this setup is that you must be effectively logged in as root and not only run script with sudo, which can work the same in this case, but also can fail, depending on how is it by default configured (and we are not checking it obviously).

You need to specify a valid url to install a site

Posted by Chipie on September 2, 2010 at 7:32pm

I was able to run BARRACUDA and OCTOPUS successfully. Thank you for your great work.

I am not able to create a site. When creating a site, I get You need to specify a valid url to install a site as an error. Any ideas?

Task starts processing
Running: /data/disk/host/tools/drush/drush.php --uri='openatrium.ecade.me' provision-save '@openatrium.ecade.me' --backend
Drush bootstrap phase : _drush_bootstrap_drush()
Found command: provision-save (commandfile=provision)
Initializing drush commandfile: drush_make
Initializing drush commandfile: drush_make_d_o
Initializing drush commandfile: provision
Load alias @self
Load alias @platform_Atrium10beta9devP001
Load alias @server_master
Loading mysql driver for the db service
Loading nginx driver for the http service
Load alias @openatrium.ecade.me
Template loaded: /data/disk/host/.drush/provision/provision_drushrc_alias.tpl.php
Changed permissions of /data/disk/host/.drush/openatrium.ecade.me.alias.drushrc.php to 600
Generated config Drush configuration file
Changed permissions of /data/disk/host/.drush/openatrium.ecade.me.alias.drushrc.php to 400
Command dispatch complete
Peak memory usage was 6.78 MB
Running: /data/disk/host/tools/drush/drush.php @openatrium.ecade.me provision-install --backend
Drush bootstrap phase : _drush_bootstrap_drush()
Load alias @openatrium.ecade.me
Drush bootstrap phase : _drush_bootstrap_drupal_root()
Loading drushrc "/data/disk/host/aegir/distro/001/drushrc.php" into "drupal" scope.
Initialized Drupal 6.19 root directory at /data/disk/host/aegir/distro/001
Found command: provision-install (commandfile=provision)
Initializing drush commandfile: drush_make
Initializing drush commandfile: drush_make_d_o
Initializing drush commandfile: provision
Trying to get property of non-object provision.environment.inc:562
Load alias @server_master
Loading mysql driver for the db service
Loading nginx driver for the http service
Including /data/disk/host/.drush/provision/db/install.provision.inc
Including /data/disk/host/.drush/provision/example/install.provision.inc
Including /data/disk/host/.drush/provision/http/install.provision.inc
Including /data/disk/host/.drush/provision/platform/install.provision.inc
You need to specify a valid url to install a site
An error occurred at function : drush_provision_drupal_provision_install_validate
Command dispatch complete
Peak memory usage was 8.88 MB
An error occurred at function : drush_hosting_task
Command dispatch complete
Peak memory usage was 21.34 MB

So Aegir thinks .me is not a

Posted by omega8cc on September 13, 2010 at 2:34am

So Aegir thinks .me is not a valid domain ext for some reason? Have you found what caused that?

Do we need to run both?

Posted by SeedTreeLLC on September 5, 2010 at 12:17am

I am confused by the two instances. If I don't need all the platforms and only need Drupal 6.x then do I need to tun Octopus script too. I don't understand why running 2 instances of Aegir is necessary. Can't you run all platforms under a single instance?

Octopus is optional; Aegir/Drush upgrade is still req'd

Posted by amstercad on September 7, 2010 at 6:24am

As I understand things, (because I haven't got Octopus to run yet), no you don't need both. Octopus is optional, but Barracuda is a req'd first step just to install Aegir hosting on a clean VPS. The Octopus script installs several different distributions of Drupal for you to use for new sites; but you can always choose to install such a distribution yourself, 'the old fashioned way'. You're still just installing Drupal in an Aegir server configuration; you just don't get the cool Octopus script from Omega8 setting the many Drupal distributions up for you en masse.

Chances are good, sooner or later you'll want to know how to install or upgrade a platform yourself anyway.

Aegir allows you to safely manage development and production code on the same server. To safely upgrade a production site when modules updates come 'round:

duplicate the codebase-platform using the command-line (and then delete the drush.rc file in Drupal-root along with any drupal/sites/example.com folder)
create/verify a new platform from it
clone the production site to be a new test site on the new platform, (with .devel somewhere in the URL if you want more than a single CSS/javascript file for inspection, firebug, etc.)
use Drush to upgrade the modules, via the test site
test to your satisfaction that all runs OK, Then you're ready to actually migrate your production site to the safely-improved platform.
backup the production site files/Db
migrate to the newly upgraded platform

Following confusion, i

Posted by a_c_m on September 7, 2010 at 9:27pm

Following confusion, i chatted with omega8 to put together this doc which spells out the above workflow in a bit more detail: http://ietherpad.com/OmbqOpxs2l

I simplified this recipe and

Posted by omega8cc on September 13, 2010 at 2:25am

I simplified this recipe and added also fast version there :)

The core Hostmaster instance

Posted by omega8cc on September 13, 2010 at 2:32am

The core Hostmaster instance created by Barracuda is for compatibility with imports from previous vanilla installs, with standard /var/aegir path, to be managed manually, so it doesn't include any platforms. The core Hostmaster manages also main Nginx configuration for all non-core instances.

The Octopus delivers much more than just non-core Hostmaster instances and ready to use platforms with re-used code. See latest changelog for Edition 0.4-HEAD-A13.A.

However if you need just one or two platforms and you prefer to add them manually, you don't need Octopus at all.

Memcache error when creating new Open Publish Site

Posted by Anonymous (not verified) on September 5, 2010 at 12:35am

Installed on vanilla Lucid Linode server without any problems (most beautiful thing). Installed all included platforms and I am able to create new sites on Acquia but not on Open Publish. I am getting the dreaded error;
"Memcache admin requires that the "cache_inc" $conf variable be set to include either memcache.inc or memcache.db.inc. Please read the installation instructions. (Currently using Memcache admin Memcache not included)".
I searched all the forums (link below) and found some info that suggest the install should work with the included global.inc.

omegac88 post concerning memcache: http://groups.drupal.org/node/73463#comment-227503

Peace,
Michael Clendening

Update to error creating site

Posted by Anonymous (not verified) on September 5, 2010 at 8:06pm

To recap a little; Vanilla install with Barracuda and Octopus script on Lucid 64bit on a Linode 768 server. DNS and host resolution is working perfectly.

I tested installing a new site with each install profile and have listed the results below.

Successful installs:

Profile..................Platform
Acquia Drupal........Acquia 6.19-svn P.001
Cocomore (CDC)....Cocomore Drupal cdc-6.19.0 P.001
Drupal Commons....Drupal Commons 1.0 P.001
Managing News.....MNews 1.0-beta11-dev P.001
Minimal................Drupal 7.0-alpha6 P.001
Pressflow.............Drupal 6.19 P.001
ProsePoint...........Prosepoint 0.35 P.001
Standard.............Drupal 7.0-alpha6 P.001
UberDrupal...........Ubercart P.001

Failed installs:

Profile..................Platform
Feature Server.......Feature Server P.001 Error #2
Open Atrium..........Atrium 1.0-beta8 P.001 Error #1
Open Atrium..........Atrium 1.0-beta9-dev P.001 Error #1
OpenPublish..........OpenPublish2.2.285 P.001 Error #1

Errors:

1- Memcache admin requires that the "cache_inc" $conf variable be set to include either memcache.inc or memcache.db.inc. Please read the installation instructions. (Currently using Memcache admin Memcache not included)

2- Output from failed command : Fatal error: Call to undefined function ctools_include() in /data/all/001/fserver/profiles/feature_server/modules/context/context.module on line 459

Peace,
Michael Clendening

Verify error

Posted by Ari Gold on September 5, 2010 at 11:58pm

I can verify error #1 when trying to create an Open Atrium site. On 64-bit ubuntu lucid 512mb RAM. Submitted issue on git.

Success (sorta) in creating site

Posted by Anonymous (not verified) on September 6, 2010 at 12:53am

I quick work around for creating an Open Publish or Atrium site is to temporarily delete the memcache directory from;
/data/all/001/o_contrib/memcache

And then add it back after the site is created. It will show up under modules once you verify the site you just created (but will not work if you are trying to use memcache as your caching engine).

Peace,
Michael Clendening

It is fixed in latest Edition

Posted by omega8cc on September 13, 2010 at 2:06am

It is fixed in latest Edition 0.4-HEAD-A13.A.

Support Update

Posted by omega8cc on September 7, 2010 at 4:40pm

Thank you all for the great feedback, bug submissions and for using this dual-core Aegir installer. With growing community of users it becomes hard to support it by just using comments in g.d.o and issues queue on GitHub. It starts to be hard to follow and hard to find and share advices and experiences, both for current and new users. This is why we plan to open very soon (a few days) dedicated place to make it all easier for you. Stay tuned! (and in the meantime we will try to do our best to answer all new questions, of course).

Does this bring us to Aegir Alpha 13?

Posted by zkrebs on September 9, 2010 at 4:10pm

If the upgrade script is run, can it switch to Aegir Alpha 13?

It is using HEAD still, since

Posted by omega8cc on September 13, 2010 at 2:05am

It is using HEAD still, since also alpha13 has critical issue with imports.

Update Edition 0.4-HEAD-A13.A

Posted by omega8cc on September 13, 2010 at 2:04am

This time it is a pretty big update. Please post issues on the GitHub queue at http://github.com/omega8cc/nginx-for-drupal/issues.

### Tue, 2010-09-12 21:50 - Edition 0.4-HEAD-A13.A

Added/Fixed: (upgrade recommended)

1.  Octopus now creates SSH/FTPS separate, non-aegir
    account for every non-core Hostmaster instance,
    with limited shell to avoid using commands
    like "drush up" since they should never be used
    on sites managed in the Aegir system.
   
2.  Octopus now by default sends a welcome e-mail
    with some useful intro information and access
    details to the address defined as _CLIENT_EMAIL.

3.  When Octopus is used the first time to create
    a non-core Hostmaster instance, it doesn't allow
    to skip installing all platforms, since it is
    recommended to add all available platforms with
    initial install, for easier re-using the code
    by next non-core Hostmaster instances.

4.  The second and all future non-core Hostmaster
    installs allow to choose one or more platforms
    or to skip adding platforms at all.

5.  Octopus by default honors initial domain used
    for the non-core Hostmaster instance on every
    upgrade to avoid mistakes with using different
    copies of the script for different non-core
    Hostmaster instances upgrades.

6.  Also Barracuda will always honor initial
    domain used for the core Hostmaster to avoid
    mistakes on upgrade when you don't use
    the original version of the script.

7.  Better checks if the script is running as root.

8.  Removed memcache module since cache is used.

9.  SMTP connection test is now optional.

10. Nginx version set to 0.8.50.

11. By defult Aegir 0.4-HEAD instead of alpha13
    is now installed to fix critical issues with
    importing sites.
    
    See also: http://drupal.org/node/907248

12. Solr and Chive are now optional (Yes/no).

13. Added optional install of Collectd monitor.

14. Fixed issue with SSL mode.

15. Better compatibility for upgrades from
    pre-Barracuda Nginx installs.
   
16. Now it doesn't cron before completing all
    install tasks to avoid breaking spinner.

17. Both Barracuda and Octopus now can better
    support re-starting stopped install/upgrade.

18. Octopus now refuses to run if defined domain
    doesn't resolve yet to the server IP address.
    
19. Octopus now refuses to run on system not
    created initially by Barracuda installer.
    
20. Custom FQDN hostname is now forced (if defined)
    in Barracuda before running DNS checks. 

21. Fix for some missing mime types in vanilla Nginx.

22. Updated versions of Open Atrium, Drupal Commons
    and Cocomore Drupal distros installed by Octopus.

23. Lowered memory defaults in the MariaDB configuration.

Linode Deb 5 Install prob

Posted by bmx269 on September 13, 2010 at 8:11am

=> Mon Sep 13 08:09:12 UTC 2010 INFO: Installing core Aegir Hostmaster instance, please wait...                                                           ==> adduser --system --group --home /var/aegir aegir: Succeeded.               ==> adduser aegir www-data: Succeeded.
==> Mon Sep 13 08:09:19 UTC 2010 INFO: Now running Aegir installer for backend and frontend, please wait...                                               ==> Mon Sep 13 08:09:28 UTC 2010 INFO: Simple check if core Aegir Hostmaster instance install is successful                                                  ==> Mon Sep 13 08:09:32 UTC 2010 FATAL ERROR: Something is wrong, core Aegir Hostmaster instance not installed - exit now

Tried it 2 times. same error. From newest Barracuda file.

Figured out all my errors

Posted by bmx269 on September 13, 2010 at 11:45pm

Figured out all my errors above, using 64bit Debian. Now I am setting up my sites again, and have a question. I use Drush a lot for updates, cache clearing, etc. What is the path I would use for that to show a valid site directory.

I used to go to /var/aegir/platform/, in there you would run a
../drush/drush --uri=xxx.xxx pm-update
for the sites I wanted to update. Now where do I go?

I also can not perform an update.php. I get an error 403 forbidden...

Thanks

UPDATED REQUEST:
Instead of posting again, I am editing the post. I think I need to have a document that shows the file / folder structure for this setup. That may help me for file placements as well. Thanks. Sorry for the many questions.

The full system path is

Posted by omega8cc on September 14, 2010 at 3:45pm

The full system path is displayed on every platform node, just use it to find the root of the platform.

You shouldn't use drush up on any live site/platform. It is a common mistake. Please follow most often used how-to: http://ietherpad.com/OmbqOpxs2l

Upgrading from single-core Barracuda

Posted by obrienmd on September 14, 2010 at 4:13am

Can I use the upgrade procedure (just run script as root again) to upgrade from an older single-core version of Barracuda from a month or so back?

Yes. But always backup

Posted by omega8cc on September 14, 2010 at 3:46pm

Yes. But always backup everything before, since it is still pre-alpha, experimental code.

Drush

Posted by bmx269 on September 14, 2010 at 4:54am

I had to run the command in the platform folder in the admin (my hostmaster user account) distro folder. The only side point is that drush is run as root, which then gives the files root privs, and can not be changed with the ftp account. I had to chown the modules folder after drush.

That answers one issue, now for the update.php issue.

The drush path.

/data/disk/admin/distro/001/pressflow# /var/aegir/drush/drush --uri=hxxp://xxx.xxx pm-update

Never ever run drush as root

Posted by omega8cc on September 14, 2010 at 3:54pm

Never ever run drush as root (or don't ask for support if you did that ;)
Never ever run drush up on any live site.
Always su -s /bin/bash - aegir to run drush on core Hostmaster site.
Always su - octopususer to run drush on non-core Hostmaster site.
You don't need all those complicated paths, there are aliases ready to use.
To use drush just type drush since it is already in your path.

If you use drush often, then

Posted by brianmercer on September 14, 2010 at 1:19pm

If you use drush often, then you should look into aliases. It's much easier to do

drush @xxx pm-update

than

drush --path=/var/aegir/platform/001/pressflow --uri=xxx.com pm-update

Create a file "/etc/drush/allsites.aliases.drushrc.php" and put this in it:

$aliases['xxx'] = array (
  'root' => '/var/aegir/platform/001/pressflow',
  'uri' => 'http://xxx.com'
);

$aliases['yyy'] = array (
  'root' => '/var/aegir/platform/001/pressflow',
  'uri' => 'http://yyy.com'
);

That will let you do use @xxx or @yyy for individual sites. Or if you update a module that you use in many sites you can do "drush @allsites updatedb" to run the update on all your sites. Or separate different classes of sites into different files.

You might also consider:

/data/disk/admin/distro/001/pressflow# drush sa xxx.com >> allsites.aliases.drushrc.php

And:

ln -s /var/aegir/drush/drush /usr/local/bin

Thanks for tip with aliases.

Posted by anantagati on September 14, 2010 at 2:27pm

Thanks for tip with aliases. Seems very useful.

All those aliases are already

Posted by omega8cc on September 14, 2010 at 3:56pm

All those aliases are already created by Aegir and by Octopus installer. Aegir by default uses aliases for everything now, for remote servers compatibility. And Drush is already by default symlinked to /usr/bin.

That's terrific, I didn't

Posted by brianmercer on September 14, 2010 at 5:39pm

That's terrific, I didn't know that it was all done by aegir. Sorry for the bad info.

So just doing "drush sa" should show a list of all the existing aliases.

What would cause this error?

Posted by bmx269 on September 16, 2010 at 7:35pm

What would cause this error? Running Drush as the Admin user, NOT Root. Running Barracuda / Octopus 13B.

Note: Updated projects can potentially break your site. It is NOT recommended to update production sites without prior testing.
Note: A backup of your package will be stored to backups directory if it is not managed by a supported version control system.
Note: If you have made any modifications to any file that belongs to one of these projects, you will have to migrate those modifications after updating.
Do you really want to continue? (y/n): y
WD php: Warning: rename(/data/disk/admin/distro/001/pressflow/sites/all/modules/mimemail,/data/disk/admin/distro/001/pressflow/backup/modules/20100916193223/mimemail):        [error]
Permission denied in drush_op() (line 597 of /var/aegir/drush/includes/drush.inc).
Failed to backup project directory /data/disk/admin/distro/001/pressflow/sites/all/modules/mimemail to                                                                         [error]
/data/disk/admin/distro/001/pressflow/backup/modules/20100916193223/mimemail
An error occurred at function : drush_pm_updatecode                                                                                                                            [error]
WD php: Warning: array_merge(): Argument #2 is not an array in _pm_update_move_files() (line 211 of /var/aegir/drush/commands/pm/updatecode.pm.inc).                           [error]
An error occurred at function : drush_pm_update     

Looks like a permission

Posted by brianmercer on September 16, 2010 at 10:25pm

Looks like a permission problem when it tries to backup a module before updating it.

I don't know the proper aegir way to do this. When I have this problem on a non-aegir system, I choose a directory to put module backups and put this directive in my drushrc.php settings file:

$options['backup-dir'] = '/some/directory/for/module/backups';

Update Edition 0.4-HEAD-A14.A

Posted by omega8cc on September 17, 2010 at 1:50pm

This update includes a few fixes for critical issues with both Aegir and latest broken Lenny installs on 32bit systems, with missing locales etc.

We would like to thank the Linode team for providing a dev VPS instance for easier and faster debugging issues for Linode users.

Please post your issues on the GitHub queue at http://github.com/omega8cc/nginx-for-drupal/issues.

Before you submit a bug, make sure you have diagnosed your configuration as documented in this guide: http://groups.drupal.org/node/21890. It is Aegir specific, but the good rules are the same: always search for similar bug report before submitting your own, and include as much information about your context as possible, especially please include the contents of files /var/aegir/config/includes/barracuda_log.txt and /data/disk/user/log/octopus_log.txt.

Enjoy!

### Fri, 2010-09-17 11:00 - Edition 0.4-HEAD-A14.A

Added/Fixed: (upgrade required)

1. Barracuda and Octopus by default installs now
   Aegir HEAD to use the fix for critical issue
   on sites import. It will be included in alpha14,
   please don't use alpha13.
   
2. Debian Lenny on 32bit systems works again.
   Fix for broken git-core after upgrade
   to version: 1:1.5.6.5-3+lenny3.1 on Lenny 32bit.

3. Fix and better inline warnings/info about
   missing locales at Linode and RackSpaceCloud.
   
4. More details in the installer log for better
   debugging and version tracking. 

5. E-mail address for alerts on database repair
   started by auto-healing now correctly replaced. 

6. Redis for Lenny now built from sources due to
   apt version moved already to Squeeze.

7. Critical bugfix for failed platforms install
   when hostmaster is not upgraded.
   
8. Introducing simple editions archive:
   http://omega8.cc/dev/bo-a14a.tar.gz 

9. Octopus now better supports using newer shared
   code for platforms and introduces new setting:
   _HOT_SAUCE to allow forced fresh/hot code.

Drush

Posted by bmx269 on September 18, 2010 at 6:09am

I have a question, I was told that I am not supposed to run Drush pm-update on my sites that are hosted with Aegir, root or admin user, or whoever. This is cool, but I can not find any information as to why, or that I was not supposed to in the first place. Can you please help, as I am new to this system, and do not want to break it. Thanks

I hope it was answered in

Posted by omega8cc on September 18, 2010 at 7:10am

I hope it was answered in this issue comments: http://github.com/omega8cc/nginx-for-drupal/issues/issue/67 especially in this comment: http://github.com/omega8cc/nginx-for-drupal/issues/issue/67#issue/67/com...

Pre-Install Error

Posted by rovo on September 22, 2010 at 3:11pm

Thanks for the awesome script. I seem to have installed Barracuda with no problem, was able to access the control panel from the browser.

I then ran Octopus, and catch a small error during pre-install

pre-install start
ERROR 1045 (28000): Access denied for user 'root'@'localhost' (using password: YES)
/usr/bin/mysqladmin: connect to server at 'localhost' failed
error: 'Access denied for user 'root'@'localhost' (using password: YES)'
==> Wed Sep 22 13:36:18 UTC 2010 UPGRADE A: pre-install completed

It says it completed, but not sure it is correct, as I am still unable to access the control panel at the subdomain I set in Octopus.sh.txt

Any advice on what I might be missing would be awesome.

*note, this is being attempted on slicehost.

Maybe there is something I need to do with conf records?

Thanks

Did you miss MAriaDB Password ?

Posted by chipway on September 23, 2010 at 7:00am

Hi,

When BARRACUDA first run, before it ends, you have to catch the MAriaDB root password. Then set it up as MariaDB root password.

I think it is possibly stored for Aegir to handle MariaDB bases management.

Omega8cc may confirm this.

MariaDB Password

Posted by rovo on September 23, 2010 at 2:00pm

Hello,

That was exactly the issue. I watched the script closely this time and caught the password on this run, and Aegir is now running at the designated domain.

Another issue I am having is, I did not receive an email, and don't know the Aegir login... Do you know if this stored somewhere?

Thank you

The standard one-time login

Posted by omega8cc on September 23, 2010 at 3:49pm

The standard one-time login link was also displayed at the end of install. You can find it in file /var/aegir/install.log (last line there).

one-time login

Posted by rovo on September 23, 2010 at 8:55pm

That is very good to know, but is that the login for the Barracuda install? The domain I set for the Octopus install is the one I am having trouble logging into.

Am, I maybe supposed to configure something from the Barracuda Hostmaster install to establish my Octopus install?

Seems like I am missing something simple here, but really close.

*when I go to the domain I set for Octopus, I see the login screen for Aegir.

Thanks for all the help!

Update Edition 0.4-HEAD-A14.B

Posted by omega8cc on September 23, 2010 at 3:51pm

### Thu, 2010-09-23 17:30 - Edition 0.4-HEAD-A14.B

Added/Fixed: (upgrade for all pre-A14.A required)

1.  Introducing default SSL Wildcard Nginx Proxy.
    Works for all sites/hostmaster instances on
    the same server and can be used also for
    encrypted connections to Chive and Collectd.
    Doesn't interfere even with SSL enabled sites
    on the same IP (with separate certs).
   
2.  The redirects are now back and enhanced.
    Fully compatible with Nginx in any combination
    with aliases and SSL settings/modes.
   
3.  Barracuda and Octopus by default installs still
    Aegir HEAD, but the latest alpha14 also works.

4.  Octopus can define its separate IP address
    if available.
   
5.  Fixed issue with too aggresive Hot Sauce check,
    causing creating not shared copies of code
    for platforms on every install or upgrade.
   
6.  Barracuda and Octopus now allows to skip DNS
    test, to make it possible to install on any
    virtualbox with dynamic DNS/IP etc. There is
    no guarantee it will work, but another switch
    is now available, if someone needs it.
   
7.  Octopus can now turn off local Memcache and Redis
    caches and switch all sites to use defined remote
    caches.
   
8.  Forced /etc/apt/sources.list rewrite also before
    the Barracuda system upgrade.

9.  Fix for the already installed and possibly broken
    git-core. 

10. Fix for Aegir sites with .info domains, the path
    alias should now work without 403 error.

chive and cpg

Posted by attiks on September 23, 2010 at 9:55pm

never mind, stupid mistake :/

collectd

Posted by attiks on September 24, 2010 at 10:29am

I played with collectd and according the their config you can config nginx support, so i edited my nginx;conf file to add it to my localhost, but collectd is complaining about the format. Anybody has experience with this?

I have this in

Posted by brianmercer on September 24, 2010 at 2:35pm

I have this in /etc/collectd/collectd.conf

<Plugin nginx>
        URL "http://localhost/stats?auto"
</Plugin>

and this in /etc/nginx/sites-enabled/default

# Listen on loopback so collectd can collect stats
server {
  listen 127.0.0.1:80;

  location /stats {
    stub_status on;
    access_log off;
  }
}

I also have php-fpm stats, but that's with 5.3.2, I don't think the 5.2.x patch has stats.

cgp plugin

Posted by attiks on September 29, 2010 at 12:32pm

There is no plugin defined for nginx, so I created nginx plugin in /var/www/cgp/plugin/nginx.php, but is still need some love

<?php

# Collectd VMem plugin

require_once 'conf/common.inc.php';
require_once 'type/GenericStacked.class.php';
require_once 'inc/collectd.inc.php';

## LAYOUT
# nginx_connections-active.rrd
# nginx_connections-reading.rrd
# nginx_connections-waiting.rrd
# nginx_connections-writing.rrd
# nginx_requests.rrd

$obj = new Type_GenericStacked($CONFIG);
$obj->width = $width;
$obj->heigth = $heigth;
switch($obj->args['type']) {
        default:
                $obj->data_sources = array('value');
                $obj->generate_colors();
                $obj->rrd_title = $obj->args['type'];
                $obj->rrd_vertical = '';
                $obj->rrd_format = '%5.1lf%s';
                break;

}

collectd_flush($obj->identifiers);
$obj->rrd_graph();

?>

I use the php-collection

Posted by brianmercer on September 29, 2010 at 2:54pm

I use the php-collection front end in the lucid collectd 4.8.2 package symlinked from /usr/share/doc/collectd-core/examples/php-collection/ and it works fine with the nginx plugin built into collectd. I haven't tried CGP.

Upgrade Failed

Posted by zkrebs on September 24, 2010 at 4:04pm

Logged the issue here: http://github.com/omega8cc/nginx-for-drupal/issues/issue/77
0.4-HEAD-A14.B

How do I fix a partially upgraded site where my cron is no longer working, yet it appears as though another host master core was installed, but left unconfigured?

Job Opportunity

Posted by R-H on September 24, 2010 at 9:33pm

I need a sys admin to help me maintain my Octopus + Barracuda server. Perfectly fine for off-site work. Please contact me here: http://ryanhanau.com/contact

Thanks

Best way to install Aegir and Virtualmin

Posted by chipway on September 27, 2010 at 8:45am

Hi,

What is the best way/process to install both Aegir BARRACUDA/OCTOPUS and Virtualmin on the same Ubuntu 8.04 Server, to get the best of Aegir and Virtualmin ?

Thanks

Hello. Take a look here:

Posted by luiginica on September 27, 2010 at 1:56pm

Hello.

Take a look here: http://groups.drupal.org/node/85184

Have a nice day, Luigi.

Thanks, this thread was

Posted by chipway on September 27, 2010 at 3:16pm

the origin of my question.

As i like the Aegir BARRACUDA/OCTOPUS with Nginx Install script, i am wandering how to install virtualmin the best way aside of them.

i tried to install virtualmin as a webmin module, but i am not satisfied of it. Or i missed some thing.

Any advice welcome.

Please don't cross post here

Posted by omega8cc on September 29, 2010 at 11:18am

Please don't cross post here when it is already discussed elsewhere: http://groups.drupal.org/node/85184#comment-305509

Updating Octopus D7

Posted by bmx269 on September 29, 2010 at 4:43am

I am wanting to update the D7 alpha to A7 on my Octopus install, and can not get it to work. I drush dl drupal-7.x into the distro folder, verified it, and then went to add a site, it did not allow it, it gave the error "Platform:No valid choices"

I would run the Octopus file again to update it, but it is still A6. Am I missing something. Thanks

Since there were known

Posted by omega8cc on September 29, 2010 at 11:16am

Since there were known problems with previous D7 alpha versions, we don't support alpha7 yet and still use alpha6 with Aegir. It requires more testing before we will add it to Octopus. I only tried it once and only "testing" install profile was available (active) to use, so there is something to debug/fix probably (again). Please try and post solution if you can.

bmx269

Posted by omega8cc on October 4, 2010 at 12:04pm

Octopus now installs by default D7 alpha7.

More octopus quirks... Maybe.

Posted by bmx269 on September 30, 2010 at 4:02am

I have to admit, it hurts me to post this, asking a million questions is irritating.

Is there a reason IMCE would not work on a site hosted by Octopus? I can upload the file, but when I add it to the content, it loses it's path. This has got me stumped.

bmx269

Posted by omega8cc on October 4, 2010 at 12:02pm

Please post some (more) details in the GitHub issue queue.

xdrago

Posted by vedat.kamer on September 30, 2010 at 8:19am

How can I disable xdrago? My scan_nginx values are like this:

$maxnumber = 200; ### 200
$critnumber = 300; ### 300
$critnumber = 800 if ($VISITOR =~ /^(66.249.)/); ### 800

But every time when I try to reach Aegir with Google Chrome 7, xdrago blocks me and I have to run iptables --flush.

I commented first 3 lines

Posted by vedat.kamer on September 30, 2010 at 2:11pm

I commented first 3 lines from crontab:

* * * * * bash /var/xdrago/second >/dev/null 2>&1

* * * * * bash /var/xdrago/Minute >/dev/null 2>&1

30 * * * * bash /var/xdrago/FireStart >/dev/null 2>&1

02 * * * * bash /var/xdrago/purge_binlogs.sh >/dev/null 2>&1
*/60 * * * * bash /var/xdrago/clear.sh >/dev/null 2>&1
15 4 * * * bash /var/xdrago/mysql_backup.sh >/dev/null 2>&1
52 * * * * bash /var/xdrago/graceful >/dev/null 2>&1

vedat.kamer

Posted by omega8cc on October 4, 2010 at 12:00pm

Just set the maxnumber and critnumber to 300 to disable the DOS prevention. You don't need (and you shouldn't) disable any cron tasks. The 800 doesn't make any sense, since it scans only the last 300 lines of the httpd server log.

This is a Breeze!! Awesome

Posted by mindtrades on September 30, 2010 at 2:01pm

This is a Breeze!! Awesome job and Thanks a lot for sharing this wonderful script to the community!!

I have a question: If i

Posted by mindtrades on October 1, 2010 at 3:41pm

I have a question:
If i install Octopus and Barracuda , and i want to host a website in root folder in nginx /var/www/nginx-default , but it gives me a 404 not found error.
Can anybody tell me how i can solve this. I changes the permission to 777 reading on internet but it does not help me.

Non-aegir nginx site

Posted by Ari Gold on October 1, 2010 at 5:36pm

Look at /var/aegir/config/server_master/nginx/vhost.d

Static website along with barracuda on the same vps

Posted by mindtrades on October 2, 2010 at 1:06pm

I did try to configure virtual host both the ways
/var/aegir/config/server_master/nginx/vhost.d as well as in /etc/nginx/ sites-available and site sites enabled. But was not able to route the static website.
The thing is that i have installed octopus and barracuda on blank vps and it is working fine, but i need to put a website on the root folder(maybe another drupal site or a static website).I have aegir installed on aegir..com. I have always used apache and am pretty new to nginx and would be grateful if someone can let me know the process in detail.
Thanks
Vin

outside aegir

Posted by rovo on October 2, 2010 at 1:42pm

Hi Vin,

I was able to install an instance of a plain site outside and away from my Aegir install. I put my site files in /var/www/sites/

Then I added a conf entry in /etc/nginx/conf.d/NEWSITE.conf

Also created an entry in /etc/nginx/sites-available/NEWSITE

This may not be completely correct, but should give you a good enough idea to get it going. You will need to restart nginx.

To best answer this question you would most likely be best to find the answer in irc #nginx

You can usually find me there as rovo

Feel free to direct message me or join this group http://groups.drupal.org/nginx for more answers.

If you start a new discussion

Posted by brianmercer on October 2, 2010 at 3:01pm

If you start a new discussion in the http://groups.drupal.org/nginx group describing exactly what you'd like to do, folks would be happy to help.

http://groups.drupal.org/node

Posted by mindtrades on October 2, 2010 at 3:58pm

http://groups.drupal.org/node/96844

done

Wildcard HTTPS Proxy question

Posted by toddgeist on October 4, 2010 at 12:48am

When I re-run the installer to do the upgrade, I get the following message

We can't enable Wildcard HTTPS Proxy for all your hosted sites,
including all Hostmaster instances, because you have already
enabled SSL on the IP address used by your core Hostmaster instance.

If you wish to use provided now by Barracuda Wildcard HTTPS Proxy,
please disable the SSL you enabled, or use the built-in Aegir SSL
feature on another IP address available and not used by the Hostmaster,
and other hosted sites

I am assuming that this is because I already have ssl and certificate installed for one of the sites in aegir, using the built in Aegir feature, and my server only has a single IP address... ?

If I want to try to use the Https Proxy, do I just need to disable SSL on the offending site and then re-run the installer.

Thanks

Todd

Posted by omega8cc on October 4, 2010 at 11:55am

Yes, you need to disable the SSL in the Aegir first, so it will no longer listen on port 443, then run Barracuda (and Octopus if you are using it) scripts, accepting upgrades of hostmaster. Then you can enable that site in Aegir with its own certificate again. The system allows you to use the wildcard HTTP proxy for all sites on your server, and still you can use the built-in in Aegir separate certificate for the site using its own cert.

How do you add Multiple Servers with Barracuda/Octopus?

Posted by Anonymous (not verified) on October 23, 2010 at 9:14am

Has anyone setup a second server with the Barracuda and Octopus script? Or what is the recommended steps to add a second remote, Barracuda/Octopus performance optimized, server to one that is has been built with these two scripts?

Peace,
Michael

Gotta it figured out...but needs clean-up

Posted by Anonymous (not verified) on October 26, 2010 at 6:07am

I got a couple of remote servers built with Barracuda and Octopus scripts nicely taken orders from a master host server. I will post the results as soon as I deconstruct the process. But first, I will send Grace a courtesy email as not to pollute the discussion with potential support request for what really is a fork to her scripts.

Peace,
Michael Clendening

Couple Of Questions

Posted by jamiet on October 27, 2010 at 2:47pm

OK well I am seriously thinking about switching to Linode from Burst.net as due to burst's non std setup Barracuda does not want to install and it looks like most people are reporting success using Linode. I have a few of questions first:

-I get the idea behind barracuda vs octopus and can buy into this approach, however I have a custom drupal platform which I create using drush make - it basically pulls the latest pressflow and the latest version of a bunch of contrib and theme modules (some module versions fixed for compatibility etc). My current workflow revolves around a central aegir setup and I run some scripts to call drush make and setup the platform - how does this fit in with barracuda and octopus? Do I run these custom scripts which saves the platform to /var/aegir/platforms/pf_20101027 and then add this as a platform to the relevant octopus instance or is there a preferred way of adding custom platforms?

-Looking at linode what is the minimum spec vps I should go for. Will this run OK on 512 linode vps - I know this is like asking how long is a piece of string but basically I want an environment where I can setup new websites with relatively low traffic and mainly anonymous visitors etc. How many sites would a 512 Linode VPS server support?

-I currently use Boost for anonymous visitors and I believe barracuda sets nginx up to support boost but how does this interact with memcache? Is memcache and redis for caching logged in users?

TIA,

JamieT

It sounds like Linode 512

Posted by brianmercer on October 27, 2010 at 2:56pm

It sounds like Linode 512 will be fine. And migrating up to another size is fairly painless.

Hello Brian, Thanks for the

Posted by jamiet on October 27, 2010 at 3:13pm

Hello Brian,

Thanks for the response. Do you currently use linode and if so what has your experience been like so far? Also if you don't mind me asking - what linode package do you have with what type of server load (drupal sites etc)?

TIA,

JamieT

I've had a 512 at the Newark

Posted by brianmercer on October 27, 2010 at 8:32pm

I've had a 512 at the Newark data center for about a year and a half. My experience has been very good. There've been a couple quick outages, but the Linode staff was always responsive, within 5-10 minutes. I never felt that their response was too slow, and that's important.

I also ran some tests on Rackspace Cloud Server and their Chicago servers gave me more php performance than Linode (better CPUs, I suppose), but the control panel, especially for DNS, was clunky, and their (supposedly godly) customer service wasn't that impressive. Besides, they only had 64-bit images which use too much extra memory for a 512M instance. I read that Slicehost (Rackspace bought them) has made some 32 bit images available.

Linode has good DNS. Five geographically distributed servers with a nice interface. Paravirt kernel so they do the kernel updates. Or you can compile your own kernel with pv-grub if really necessary. Alert notifications for bandwidth or CPU use. Cloning and resizing. Cheap additional IP addresses for ssl. ($1 a month instead of $2-3 like most places)

I haven't used their new backup system. I use a duplicity cronjob to amazon s3.

I don't have high load sites. Just a few small communities and semi-static sites. All Drupal. I use Boost where possible, some use forums so a chunk of the traffic is authenticated. The sites remain responsive at all times of day.

I use an nginx/php-fpm setup like octopus/barracuda which is pretty efficient and perfect for a low memory server. My collectd graphs show minimal cpu-steal.

Now that you forced me, here's my referral link http://www.linode.com/?r=f0b45b52df6544ea2ec664c921daf0c516b6a05a Omega8cc may have a referral link somewhere and since she does all the work on the octopus/barracuda script, you might want to use her referral link instead. They don't give you anything, just some credit for us.

You can add your custom

Posted by omega8cc on October 27, 2010 at 9:03pm

You can add your custom platforms to the Hostmaster core instance created by Barracuda or to any satellite Hostmaster instance created by Octopus. Just remember to use correct system user, since every instance has its own user (the core Hostmaster is using standard "aegir" user, of course).

It doesn't make any difference from the Barracuda/Octopus perspective how are you building your custom platform. It will be always compatible, so long as your custom platform will work with the Aegir verify/provisioning standard system.

Memcache and Redis are configured to speed up things only for logged in users by default - you just need to enable the "cache" module on any site created in the Octopus instance. Of course you have to add this module also to your custom platform to be able to use it.

I believe Brian explained all other details perfectly and furthermore he is the author of the Boost compatible original Nginx configuration we are using, he helped a lot with promoting Nginx in English, so he really deserves the "thank you" for his work. If you plan to use Linode, please use Brian's referral link.

By the way, since many people asked me already if they can donate money to this project development, and since it is already supported by my company, I will probably open a ChipIn account for those interested and all funds will be forwarded to some good charity.

After turning on the cache

Posted by obrienmd on October 27, 2010 at 10:35pm

After turning on the cache module (6.x-1.0-beta1 will work I assume), do we need to do any of the customization to settings.php as noted on the cache module page?

How about boost? Any configuration needed (UI or conf files), or just turn it on and go? Are there benefits to using boost on top on nginx?

Would love more information about using these modules for improving the already very good performance of this script!

Option to wipe it out

Posted by kunago on October 30, 2010 at 3:33pm

I am just trying to put up with my first BARRACUDA installation on a virtual machine in order to test it before I will go for a VPS and have been thinking whether it is currently possible to uninstall BARRACUDA/OCTOPUS completely from a server? I am not a linux power-user so please bear with this stupid question.

It could but necessarily is not useful. Just a question for a discussion.

It's easy

Posted by skwashd on November 4, 2010 at 11:19am

I am not sure if you want to keep Aegir and just drop the bits barracuda installs or if you wish to drop it all. I will explain how to drop it all.

To ditch Aegir, first delete all the sites and platforms you have added. Next you need to delete Aegir itself, so just remove the /data directory. Finally drop the Aegir database.

I am not sure what other changes barracuda makes to a server when it is being installed, so you might have to pick through the scripts to see if there is anything else relevant that needs to be rolled back.

To clarify it a bit:

Posted by omega8cc on November 4, 2010 at 1:21pm

To clarify it a bit: Barracuda will install complete environment to run Aegir plus the core (main) Hostmaster instance in the standard /var/aegir directory.

Every Octopus instance is created in /data/disk/foo subdirectory.

I will release also the basic Barracuda version to create just LEMP/LXMP environment, so you will be able to use it for the remote web/db head or for manual Aegir install.

There is no reason to uninstall what Barracuda will install for you. If you need to start fresh, just re-image the server/vps/vm completely.

If you want to re-install just core (Barracuda) and/or satellite (Octopus) Hostmaster instance, you should (as skwashd already explained) delete all sites first (to purge no longer used databases), then delete Aegir system users (deluser aegir and deluser octopususer) and finally delete the /var/aegir and /data directories.

Now you can re-run Barracuda to install Aegir again, without going with full system re-install.

Security

Posted by zkrebs on December 13, 2010 at 12:44am

Omega8cc - I just stumbled across this article, top 20 ways to harden the nginx server. How many of these are being applied in Barracuda's setup? I just saw that Gawker was compromised...led me to re-evaluate the security of my setup.

http://www.cyberciti.biz/tips/linux-unix-bsd-nginx-webserver-security.html

I believe Barracuda installs

Posted by omega8cc on January 20, 2011 at 8:06am

I believe Barracuda installs enough secure environment out of the box. Also, every good sysadmin should use its own best practices for high security so not all possible tweaks/settings are published/included. It is your responsibility, while Barracuda gives a good start out of the box. Also, some tips from that article are far too paranoid and would break most of "normal" production environments so not all are useful.

After upgrade I get install.php

Posted by toddgeist on January 7, 2011 at 6:02am

I just re-ran Barracuda and Octopus on my functioning server, to do an upgrade. Everything went see,ed to go fine, but when I visit my Octopus site at "host.mdomain.com" . (host is the user for my octopus install) I am taken to "install.php". and asked to choose an install profile. Either Hostmaster or Drupal.

So something seems weird.

further examination shows that my host install was moved to the 003 distro

/data/disk/host/aegir/distro/003/

but in the browser I am still being directed to

/data/disk/host/aegir/distro/002/

which is empty and there for giving me an install page.

Can anybody help me sort out what went wrong? and how to fix it?

Thanks

Todd

fastcgi_params.conf was missing

Posted by toddgeist on January 7, 2011 at 2:09pm

Starting and stopping nginx turned up an an issue.

Starting Nginx Server...
nginx: [emerg] open() "/data/disk/host/config/includes/fastcgi_params.conf" failed (2: No such file or directory) in /data/disk/host/config/server_master/nginx/vhost.d/host.mydomain.com:3

i removed the include and nginx started up again. But don't i need that fastcgi_params.conf to be included some where?

Thanks

Todd

Tuning for low RAM

Posted by dev286 on January 12, 2011 at 9:59am

First, great job on Barracuda and Octopus... Super fast! Everything works great so far.

I experience some performance issues... My setup is on Linode 512, Ubuntu 10.04 LTS, basic Drupal platforms. After boot, the whole thing runs really fast, but after it begins to hit virtual memory it slows down... just a little, maybe 5-10 seconds. I guess this is due to the virtual memory access time on the server. Can you suggest any performance tweaks for low RAM setups.

Thanx
dev286

First I'd like to thank you

Posted by smira on January 20, 2011 at 5:36am

First I'd like to thank you for the great work here to all that contributed!

Secondly, I too am wondering if there are any tweaks for those of us running with low RAM ?

Any help would be appreciated.

Thanks again!
-smiro

Thanks! The question about

Posted by omega8cc on January 20, 2011 at 7:58am

Thanks! The question about running Barracuda and Octopus with low memory has been already answered here: http://groups.drupal.org/node/84074#comment-382879

Fresh Install fail

Posted by dev286 on January 12, 2011 at 8:46pm

Weird install failed..... on Ubuntu 10.04 LTS

Do you really want to proceed with the install (y/n): y
Project drush_make (6.x-2.0-beta9) downloaded to /var/aegir/.drush//drush_make. [1;32;40m[1m[success][0m
Project drush_make contains 0 modules: .
Command provision-install needs a higher bootstrap level to run - you will need invoke drush from a more functional [31;40m[1m[error][0m
Drupal environment to run this command.
The drush command 'provision-install' could not be executed. [31;40m[1m[error][0m
A Drupal installation directory could not be found [31;40m[1m[error][0m

Any ideas?

Please follow bug/problems

Posted by omega8cc on January 20, 2011 at 8:00am

Please follow bug/problems submission guidelines (see README). We don't use g.d.o for bugs/problems submissions. There is an issue queue on the GitHub.

Understanding Performance modules in Octopus

Posted by tribe_of_dan on February 24, 2011 at 2:22pm

1st up, Octopus is amazing... I'm so grateful for all the work you guys have put into this!

I wonder if you can help shed some light on the performance modules for me. I've installed OCTOPUS on a Linode VPS and have a few questions.

Cache Module - is it as simple as enabling the module? I noticed Cache is a part of the Pressflow Platform.

On the Cache project page (http://drupal.org/project/cache) under Installation, it says "2. SETUP WEB-SITE'S settings.php CONFIGURATION FILE". I keep getting the WSOD each time I edit settings.php. I think my path to cache.inc is wrong.

What should the correct path be for an Octopus / Pressflow 6 install?

Also, after I enable Boost module I'm told... "Boost successfully installed. Fix all errors and warnings on the status report page under Boost for a quick installation."

On my site status report it says:

APC stat files Enabled
APC stats PHP files to determine if they have been updated when apc.stat is set to 1. Disabling updates (set to 0) will increase performance. But web-server restart or call to apc_cache_clear() will be required to update.

Cache Not configured.

How would one go about fixing these issues? I'm using Pressflow Drupal6. Thanks.

You only need to enable

Posted by omega8cc on March 11, 2011 at 12:12am

You only need to enable performance related modules. The rest is already done for you. It is also explained in the welcome e-mail you will receive after creating Octopus instance. The issue with "Cache not configured" is fixed in head/master and the fix will be included in the upcoming (I hope today!) 1.0-boa-T release.

Perfect! Thanks so much :)

Posted by tribe_of_dan on March 11, 2011 at 12:18am

Perfect! Thanks so much :)

Looks incredible

Posted by terkhorn on March 10, 2011 at 11:47pm

I am totally hyped to play with this. Is it too early to say that you rock? PROBABLY NOT!

-f

--
Felix Terkhorn
If it were easy to understand, it wouldn't be called code.

Thanks Felix!

Posted by omega8cc on March 11, 2011 at 12:01am

~Grace

"looking for the best/fastest Ubuntu mirror"

Posted by terkhorn on March 17, 2011 at 6:54pm

Hi Omega8cc,

I'm guessing this step shouldn't take two hours to complete ... for the Barracuda install, I made it to the INSTALL START -> checkpoint: and I hang at

Barracuda [Thu Mar 17 12:53:40 EDT 2011] ==> INFO: Now looking for the best/fastest Ubuntu mirror, it can take a while, please wait...

This is on a fresh Ubuntu 10.10. I watched the network traffic via netstat when restarting this once, and saw the process start polling very many servers (~30), then eventually it filtered down to just holding a wget connection open to linux3.cc.ntu.edu.t:www in state 'ESTABLISHED' for the next couple of hours.

If you'd prefer I put this in the issue queue, let me know?

Thanks!
Felix

--
Felix Terkhorn
If it were easy to understand, it wouldn't be called code.

moved to github issues queue

Posted by terkhorn on March 17, 2011 at 7:42pm

I did post this in the github issues queue for tracking purposes -- should have gone there originally. Thanks,
-f

--
Felix Terkhorn
If it were easy to understand, it wouldn't be called code.

Positive feedback exists inside this comment

Posted by terkhorn on March 24, 2011 at 5:18pm

Hi all,

I just want to report that after switching my host to Debian 6.0.0 as recommended by the install guide, and working through some minor issues, this is really a great setup. I'm now migrating sites from an old host into this container, as it looks like a great way to have some point-and-click manipulation of a dev & production environment. Kudos, kudos, kudos to omega8cc for making a great product available here!

I have a lot of experience with apache but didn't have time to tackle even a small learning curve for nginx due to production deadlines at my job. This made the transition painless and will allow me to come up to speed on nginx on my own time. If you have the ability to procure a Debian shard, give this a try!

Cheers,
Felix

--
Felix Terkhorn
If it were easy to understand, it wouldn't be called code.

FQDN

Posted by emackn on April 5, 2011 at 12:17am

I dont know what it is, I think I have the FQDN set up correctly.. but I continually get the "Invalid DNS setup" message.

This is on a fresh install of ubuntu-server 10.10:

Here's the output of all the commands I can think of to debug this.

root@aegir:~# hostname
aegir.elocal
root@aegir:~# hostname -f
*.aegir.elocal

root@aegir:~# more /etc/hosts
127.0.0.1 localhost
192.168.74.194 *.aegir.elocal aegir.elocal aegir

The following lines are desirable for IPv6 capable hosts

::1 localhost ip6-localhost ip6-loopback
fe00::0 ip6-localnet
ff00::0 ip6-mcastprefix
ff02::1 ip6-allnodes
ff02::2 ip6-allrouters

root@aegir:~# ping aegir.elocal
PING *.aegir.elocal (192.168.74.194) 56(84) bytes of data.
64 bytes from *.aegir.elocal (192.168.74.194): icmp_req=1 ttl=64 time=0.027 ms

root@aegir:~# nslookup aegir.elocal
Server: 68.87.73.242
Address: 68.87.73.242#53

** server can't find aegir.elocal: NXDOMAIN
*** possibly my issue.

root@aegir:~# more /etc/resolv.conf

Dynamic resolv.conf(5) file for glibc resolver(3) generated by resolvconf(8)

DO NOT EDIT THIS FILE BY HAND -- YOUR CHANGES WILL BE OVERWRITTEN

nameserver 68.87.73.242
nameserver 68.87.71.226
search elocal

root@aegir:~# uname -n
aegir.elocal

Been one of those days, nothing seems to work correctly. What am i missing here?

Thanks,
eric.

http://www.twitter.com/ericmckenna

root/.workspace/.garbage.

FQDN from host suggestion

Posted by JeffBrower on April 5, 2011 at 1:30am

It looks like it is getting the resolution from the hosts for your ping and not the dns - which is fine. That brings you back to the possibility that DNS is actually being an issue.

Just throwing stuff against the wall here, but: Since this appears to be an internal LAN domain (elocal), can you walk up to the DNS server and see if it actually resolves there? It also looks like this is set up as a DHCP and gets clobbered with every lease which could cause you issues. Are you SURE the FQDN is correctly entered in both places - I have seen things like e.local and elocal.aegir.mydomain.com and such and confusion happens. Last thing I can think of off the top of my head is are you sure the DNS server was restarted after the configuration change and if you are hitting slaves, were the serial numbers on the masters incremented so the slaves knew to reload their zone files (generally speaking in bind9 servers they need that rather than just a restart).

I guess the other thing that looks different from what I do is that I have a single name in the hostname like "aegir" and in the /etc/resolv.conf file I would have both the domain and the search set up (if I used elocal) like:
search elocal
domain elocal
nameserver 68.87.73.242
nameserver 68.87.71.226

Then I can do the dns resolution with both aegir and aegir.elocal using dns.

I also don't use wildcards in the hosts file with aegir since it modifies the host file, but that's just me.

If the machine looks right, the simplest answer is that it is really telling you about a real issue. Maybe it's really not you. :)

I hope that something there leads you to something that leads to something that helps. I know how a day like this can be.

In Barracuda there are three

Posted by omega8cc on May 1, 2011 at 10:21pm

In Barracuda there are three options:

You can force your local settings in _MY_OWNIP, _MY_HOSTN and _MY_FRONT
You can use the trick with _LOCAL_NETWORK_IP and _LOCAL_NETWORK_HN
You can disable the FQDN check to see if this will work anyway.

In Octopus you can force your local IP with _MY_OWNIP.

You can also find many helpful tips in our issue queue at https://github.com/omega8cc/nginx-for-drupal/issues

@emackn

Posted by omega8cc on August 26, 2011 at 9:14am

After reading this again, I think the problem is that hostname -f returns *.aegir.elocal. You can use wildcard in the dns entries, but not in the FQDN hostname. It should be anything.aegir.elocal, IMO.

Access denied

Posted by denny84 on June 5, 2011 at 2:07am

I am installing aegir on virtualbox but after installing everything, when I got my local.dev server (this is where I have installed aegir) I get Access Denied. Also I dont know the username and password to login into my Aegir. Where could I go wrong? I have set up the dnsmasq and all the stuff mentioned above but somewhere, something is wrong. I have not installed any profile through octopus as I dont know if I did something wrong. I could not find any answer on git repo so I am posting here.

Thanks a lot!!

login

Posted by diogogomes77 on August 19, 2011 at 2:01pm

the link for setting admin passwd can be found at the end of /var/aegir/install.log

Accessing Octopus

Posted by diogogomes77 on August 19, 2011 at 2:41pm

Hello I'm new at this and very excited with the Aegir concept.

I've a fresh install of Ubuntu-server 10.04 LTS
my hosts file:

168.0.14    aegir.mydomain.info
0.0.1 localhost
0.0.1 aegir.mydomain.info aegir   mydomain.info   
0.0.1  satellite.aegir.mydomain.info   satellite

hostname: aegir.mydomain.info

then BARRACUDA executed well with these settings and control pannel is running on aegir.mydomain.info

_DNS_SETUP_TEST=NO
_AEGIR_VERSION=1.0-boa-T-8.9
_MY_OWNIP="192.168.0.14"
_MY_HOSTN="aegir.mydomain.info"
_MY_FRONT="aegir.mydomain.info"

Then executed OCTOPUS with _DNS_SETUP_TEST=NO and the install ended sending me an email with all the credentials and links to access the created sattelite.

The FTPS/SSH account works well, but http access to host: satellite.aegir.mydomain.info just shows the "Under Contstruction" Nginx's default server

Lots of files and folders were created on /data/ and /home/ so I guess It's all there. It's just the DNS/Nginx that's not working, right?

in /var/aegir/config/server_master/nginx/platform.d there's a symlink:
include /data/disk/satellite/config/server_master/nginx/vhost.d/*;
but that folder is just empty :(

then in /var/aegir/config/server_master/nginx/vhost.d there's the vhosts for hostmaster and chive

so there's no more connections to the satellite Octopus installed ?

could I create a vhost manualy so it points to the sattelite? What would be the root path for those files?

Thanks in advance

It's working now!

Posted by diogogomes77 on August 19, 2011 at 5:55pm

After all, the Octupus installation was not 100%, there was a problem accessing mysql during install.

Had to run the script once again from start choosing a different subdomain.

Now that I have access to the "cockpit" let's have some fun :D

What if?

Posted by cwittusen on August 26, 2011 at 5:15pm

I already have nginx, mysql, php-fpm, memcache installed; do I need to remove all that before I do this instead? I like this because not only is it nginx but also creates a more centralized configuration than what I currently have and I'm all for that so this is a great solution.

Please let me know if I can run this install without removing my current install and functions and once I have tested it and ported the other sites over to this new setup then I can remove the old stuff; appreciate any and all responses.

Regards,
Chris W

I don't recommend running

Posted by omega8cc on August 29, 2011 at 12:48pm

I don't recommend running Barracuda install on any live system. Also, we require minimal OS install without any services installed (Postfix or Exim are often bundled by default, but it is OK). Of course you could try this, but then be prepared for disaster recovery, as it may overwrite your existing settings and simply take down your running sites. It may be not that bad, however there is no guarantee of any kind.

Re: I don't recommend running

Posted by cwittusen on August 29, 2011 at 2:23pm

So I can re-image my server and request that its a standard Ubuntu install without apache, mysql, postfix and exim; then do I have to install barracuda, octopus and boa first before running your script or does it all comes with the install script?

Is it possible to ask for a bullet list of how to run this install? I saw one earlier but its a bit old and there seems to have been some changes to that way of running it. Finally, I have about 10 domains hosted on my server but the DNS hosting is via godaddy; however, to setup these 10 sites, do I have to run the script 10 times (1 for each domain) or is there another way of doing it?

Thank you for all your help.

Chris

Regards,
Chris W

@wittusen

Posted by omega8cc on August 29, 2011 at 2:48pm

Please open a support request in the Barracuda issue queue or create a new thread in our BOA group if you wish to discuss this with community.

Aegir install

Posted by cwittusen on August 29, 2011 at 4:26pm

I thought I was discussing this with the community; I was just looking to find out if I actually have to install barracuda, octopus and boa first before running the install script that I think you have built.

What other community do I need to turn to in order to get that question answered?

DISREGARD THIS POST, I FOUND THE BOA LINK.

/Chris

Regards,
Chris W

This is an old and in fact

Posted by omega8cc on August 29, 2011 at 4:28pm

This is an old and in fact outdated already thread created when there was no BOA group yet.

Now it is better to start new thread in our group at http://groups.drupal.org/boa

Thank you

Posted by cwittusen on August 29, 2011 at 4:37pm

Thank you; I found the link to the BOA page and I know am getting acquinted with it and then I will go ahead and install B first and The O and then have myself a BOA. :)

chris

Regards,
Chris W

Client can't access site -- firewall has been disabled

Posted by terkhorn on August 29, 2011 at 2:30pm

Hi,

Bit of a strange question -- and not necessarily one that is due to nginx/barracuda/aegir, but which might be related.

I delivered the first draft of a site to a client and found that LFD+CSF were blocking his IP, as well as mine at home (but not mine at my workplace).

I've done /etc/init.d/lfd stop and /etc/init.d/csf stop, and now several friends can see the site -- actually, everyone who isn't the client can see it -- but the client still can't connect to the server.

I'm watching /var/log/nginx/access.log and I'm mostly seeing "200 status" messages, though I'm occasionally seeing some that look like this, and they come in sets of 5:
_ [29/Aug/2011:09:53:00 -0400] "-" 400 0 0 0 "-" "-" 0.000 "-"

I'm also watching /var/log/messages and that seems pretty sane. Earlier I was seeing "TCP IN" firewall blocking messages, which have gone away since I shut down CSF+LFD.

Can you give me any suggestions on how to diagnose why my client isn't seeing the site? He's cleared his cache, and tried from two different browsers. Maybe there are some clues in logs that I haven't thought to check yet.

Also...
One more thing. The client may have a gummed-up DNS experience as the site is currently attached to a C NAME record, which may be conflicting with his own network settings. Is it safe for me to run BARRACUDA to reset my FQDN, so that I can serve this from a different DNS domain? He might be able to get through to the site, then.

Overall -- a very positive experience -- when my site actually can be accessed, it's performing quite quickly, even with drupal's cache turned off and "theme development mode" enabled. bravo!

Cheers,
Felix

--
Felix Terkhorn
If it were easy to understand, it wouldn't be called code.

@terkhorn

Posted by omega8cc on August 29, 2011 at 2:46pm

Please don't post issues/support requests as comments in the very old g.d.o thread.

It is also explained in the BOA group header at http://groups.drupal.org/boa

thanks

Posted by terkhorn on August 29, 2011 at 2:57pm

moved to barracuda issues queue

--
Felix Terkhorn
If it were easy to understand, it wouldn't be called code.

Remote Servers

Posted by dev286 on September 13, 2011 at 4:40am

Once again, great work! My install been working for over a year now on Linode 1Gig (with updates) - no problems!!!!

Q... How do I setup remote servers with Barracuda?

Unfortunately if you are

Posted by snlnz on September 13, 2011 at 4:47am

Unfortunately if you are trying to setup remote aegir servers this feature isn't yet implemented.

Aegir currently supports headless remote servers so you can push sites to a non aegir based apache server for example but not back the other way, so proper remote operations are not actually working yet.

It's a real shame as the dev/stage/prod workflow is still limited because of this lack of functionality but it is just how the aegir project is shaped currently to the best of my knowledge.

Octopus might be on the cards to balance mutli aegir instances but I haven't had time test it's limitations to date.
There is a bit of discussion on the aegir community site but I'd be interested to know what findings you come across if it differs from my reply. :)