We're pleased to announce new AWS AMIs for Mercury 1.1 on Ubuntu Lucid. These Beta AMIs were made using the instructions posted here: http://groups.drupal.org/node/70268. AMI IDs after the jump...
us-east:
ami-294aa340 pantheon-amis-us-east/mercury-pressflow-drupal-lucid-32-1.1-Beta2.manifest.xml
ami-554aa33c pantheon-amis-us-east/mercury-pressflow-drupal-lucid-64-1.1-Beta2.manifest.xml
us-west:
ami-2b8ddc6e pantheon-amis-us-west/mercury-pressflow-drupal-lucid-32-1.1-Beta2.manifest.xml
ami-298ddc6c pantheon-amis-us-west/mercury-pressflow-drupal-lucid-64-1.1-Beta2.manifest.xml
eu-west:
ami-6d3c1619 pantheon-amis-eu-west/mercury-pressflow-drupal-lucid-32-1.1-Beta2.manifest.xml
ami-673c1613 pantheon-amis-eu-west/mercury-pressflow-drupal-lucid-64-1.1-Beta2.manifest.xml
ap-southeast:
ami-27e39c75 pantheon-amis-ap-southeast/mercury-pressflow-drupal-lucid-32-1.1-Beta2.manifest.xml
ami-25e39c77 pantheon-amis-ap-southeast/mercury-pressflow-drupal-lucid-64-1.1-Beta2.manifest.xml
A couple of notes:
1) These are our first Mercury images to support multi-site ApacheSolr and Memcached. We'll post instructions here soon on how to set this up. Multi-site instructions are here: http://groups.drupal.org/node/72488
2) These images were made using the code at http://launchpad.net/pantheon/1.1 - this is the same path we'll use for Mercury 1.1 when it's out of Beta. This means that a simple "sudo /usr/local/bin/update_mercury.sh" will update a server made from the above images to Mercury 1.1 stable (when it's released).
3) We'll targeting Ubuntu Lucid and CentOS 5.5 (coming very soon) for Mercury 1.1. While we've done nothing (that we know of) to prevent Ubuntu Jaunty and Karmic servers from running Mercury 1.1, we strongly suggest that folks use Ubuntu Lucid (being a a long-term support release) or CentOS as they are the focus of our development and testing.
4) I have previously indicated the I thought we would release an EBS-boot AMI for Mercury. This no longer seems likely. The issue is spending development time for a platform that is so unique. However, all our code is posted at http://launchpad.net/pantheon and we would be more than happy to integrate any working code that helps folks to use Mercury on EBS-instances. Having said that, the Mercury 1.1 instructions (http://groups.drupal.org/node/70268) should install on an EBS-boot instance, but I would use "sudo bcfg2 -vqed" rather than "sudo bcfg2 -vqed -p 'mercury-aws'" in step 5 to prevent BCFG2 from moving data into /mnt (which is non-persistent on an EBS-boot instance). Then, after Mercury 1.1 is installed, one would want to install some of the AWS-specific packages like ec2-ami-tools, ec2-api-tools, and ec2-consistent-snapshot.
As always with our releases, we are very interested in feedback. Please post here any issues/questions/suggestions you have.
Thanks and enjoy!
Greg
Comments
Will the above work for
Will the above work for CentOS 5 or just 5.5?
Also, any idea of when this may be available?
re: Will the above work for
We don't have CentOS images or instructions yet, but we are working hard on both.
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
Any update on getting CentOS Images & Instructions?
Hi Greg,
Great work with Mercury!
Just wanted to follow up on your post;
have you released a CentOS image for Mercury yet?
If not, are there equivalent instructions for CentOS availalbe somewhere?
Thanks.
re: Any update on getting CentOS Images & Instructions?
Sorry - I should have updated this page. We don't have images for CentOS yet, and we won't until after we release Pantheon 1.2.
However, we do have instructions for building Pantheon 1.1 on CentOS: http://groups.drupal.org/node/80244
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
Thanks. Keep us posted. Keen to test your AMIs
Thanks for the update Greg,
Will follow the instructions for now.
Will the AMIs be using EBS when you do release?
:-)
No EBS?!?
Hey Greg,
Thanks for the update. But I must admit that I'm disappointed to learn that you're not planning to release an EBS boot image. That was the one thing I was waiting for before getting involved. I know several others who were waiting for that too.
Is the development time required to support an EBS boot image that massive? Is the issue that each time you release an AMI , you would have to go through an additional EBS prep process? Could the community do something to help?
I'm just hoping there's some type of "middle way" here...
Thanks,
Ben
The reason
It's actually really easy, which is why we're not doing it at the moment. People who want to host on EC2 should be comfortable managing Amazon's services themselves, because once you get beyond a proof of concept, how you use these services is determined by your use-case; there's not any real "one size fits all" option here.
Releasing an EBS backed AMI wouldn't be too hard (though it doubles the time it takes for us to do a release), but it also wouldn't provide very much value. If you're unsure of using EC2 because of the lack of non-volatile storage, I'd recommend either do the dive and understand EBS yourself, or work through a different hosting provider with a more traditional setup.
https://pantheon.io | http://www.chapterthree.com | https://www.outlandishjosh.com
EBS
I would definitely put in a vote for EBS. Totally appreciate the great work to date. Benefit of EBS for one is the cost savings. Ability to turn off your instance without having terminate. That seems like a killer feature.
Mercury 1.1 EBS on Lucid Test Site
Followed the Mercury 1.1 directions [ http://groups.drupal.org/node/70268 ] using the http://alestic.com/ Lucid EBS AMI. Had to add a /data folder to Solr. Just starting to test. We may just want to let Greg focus on stabilizing 1.1 and roll our own EBS AMI.
re: Mercury 1.1 EBS on Lucid Test Site
jamesJonas,
Thanks for the heads up - I had forgotten to update http://groups.drupal.org/node/70268 with the new solr path changes (made to support multi-site). I just fixed that.
Please do let us know how installing on EBS goes (and anything you had to do different).
Thanks!
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
Setting up EBS Volume
EBS volume setup Directions: http://developer.amazonwebservices.com/connect/entry.jspa?externalID=1663
Mercury on EBS: http://groups.drupal.org/node/33092
re: Setting up EBS Volume
Added the EBS volume setup Directions node to the Documentation tab.
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
EBS Volume Setup
You may wish to hold off on the linking the directions until I finish a bit more testing. Using an EBS AMI creates a 15 GB EBS snapshot volume. This should mean that nothing further is necessary, like creating a /vol XFS directory (as per the attached instructions). Right now I'm testing the EBS Start/Stop inside of Amazon to see if everything works. At the moment I'm running into a bit of trouble getting Mysql to properly restart after I have stopped the server (using Amazon native facility) and then restarted it.
I should have most likely not have posted the directions until I was clearer as to there intended use.
JJ
re: EBS Volume Setup
I meant to say that I had linked to the "Mercury on EBS" wiki rather than the "EBS volume setup Directions" page (unable to edit that post for some reason). Either way, I will remove it from the documentation tab if you like, but my suggestion is to leave it there and let others help you with the wiki.
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
Perfect Context - Documentation to a Wiki
Greg,
As Documentation, both links are relevant. The "EBS volume setup Directions" is the source document for the "Mercury on EBS". I was concerned that someone would assume that using an EBS AMI and then following the Mercury 1.1 recipe required some additional steps. As Documentation, these links are relevant as they give important background on how EBS works.
Your first intuition was the correct one.
James
Mercury 1.1 EBS tomcat restart
Greg,
I had to restart tomcat in order to get solr running on the EBS Mercury Lucid. This was most likely related to the need to add the /data folder.
If I do build an AMI for EBS, can I use the word 'mercury' and/or 'pantheon' in the AWI name:
Your naming convention:
pantheon-amis-eu-west/mercury-pressflow-drupal-lucid-64-1.1-Beta2
Example of a different naming convention:
somewebsitename-amis-eu-west//mercury-pressflow-drupal-lucid-64-1.1-Beta2-1234567890
My sense is: mercury (ok), pantheon - (nope)
James Jonas
Tomcat restart now in init.sh
James,
FYI: We just filed and fixed https://bugs.launchpad.net/pantheon/+bug/588814, which adds a tomcat restart to the init.sh script. If you run init.sh after you configure your xml file, you shouldn't have to restart tomcat manually anymore.
Greg will have to answer your other questions.
Justin
re: Mercury 1.1 EBS tomcat restart
Thanks for the report - we did find a bug in cases were the system wasn't restarted after installing Mercury (and init.sh was run instead). This has been fixed (by adding a tomcat restart to the /etc/mercury/init.sh file) and released.
As for releasing your own AMIs with Mercury, your sense is correct - you are welcome to use the mercury name but pantheon is trademarked.
Thanks and please announce here when you release the AMIs.
Thanks!
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
Did anyone release an EBS boot?
If so, can you please let me know the name used.
Thanks!
EBS Boot AWS AMI
I have done some initial testing with mixed results. Which means I have taken an EBS Boot AMI, followed the directions from Greg and spun up a server. Ran into some issues with MySQL failing to reboot. Killed that server. Hope to free up some time soon to make another attempt.
James
glibc libraries issue
On the eu-west:ami-6d3c1619 AMI, I get the following error in /var/log/syslog.
Currently emulating unsupported memory accesses in /lib/tls glibc libraries. The emulation is slow. To ensure full performance you should install a 'xen-friendly' (nosegneg) version of the library, or disable tls support by executing the following as root mv /lib/tls /lib/tls.disabled. Offending process: apache2
You can check whether you are using nosegneg version by issuing
sudo ldd /bin/truelinux-gate.so.1 => (0xf57fe000)
libc.so.6 => /lib/i686/nosegneg/libc.so.6 (0xb7e90000)
/lib/ld-linux.so.2 (0xb7fe4000)
If libc.so.6 is not pointing to the nosegneg version you can correct this by issuing the following:
sudo aptitude remove libc6-i686 ubuntu-minimalsudo aptitude install ibc6-xen
sudo echo 'hwcap 1 nosegneg' > /etc/ld.so.conf.d/libc6-xen.conf
sudo /sbin/ldconfig -v
Documentation elsewhere mentions lib6-xen conflicts with libc6-i686, therefore it and ubuntu-minimal, which depends on lib6-i686 needs to be removed first. If the kernel version (uname -r) is older than 2.16.24 then hwcap should 0 and not 1. However, the AMI´s use a much later kernel version.
Rerun
sudo ldd /bin/truelinux-gate.so.1 => (0xf57fe000)
libc.so.6 => /lib/i686/nosegneg/libc.so.6 (0xb7e90000)
/lib/ld-linux.so.2 (0xb7fe4000)
It should now point to the nosegneg version.
You might want to reboot to make sure the original error is no longer there by looking at /var/log/syslog after a reboot
re: glibc libraries issue
I check all our 32 bit AMIs and they all have this error in syslog. Our 64 but AMIs don't.
One question I have is this:
A new 32-bit Ubuntu Lucid server says:
sudo ldd /bin/truelinux-gate.so.1 => (0xb7782000)
libc.so.6 => /lib/tls/i686/nosegneg/libc.so.6 (0xb761c000)
/lib/ld-linux.so.2 (0xb7783000)
A 32-bit Mercury 1.1 Beta server Lucid also says:
sudo ldd /bin/truelinux-gate.so.1 => (0xb778a000)
libc.so.6 => /lib/tls/i686/nosegneg/libc.so.6 (0xb7621000)
/lib/ld-linux.so.2 (0xb778b000)
Doesn't that mean we shouldn't be getting the error?
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
GLIBC solutions not working
Hi Greg,
Apologies for the late response.
I used the eu-west:ami-6d3c1619 AMIi and "ldd /bin/true" did not point to the nosegneg libc version.
However, what´s more strange is that none of the workarounds that I had given above are actually working. On reboot, I still get the error in syslog. Probably a Lucid bug that´s still not caught in Ubuntu Forums.
I´ve decided to ignore that for the moment.
re: GLIBC solutions not working
per https://bugs.launchpad.net/ubuntu/+source/eglibc/+bug/427288, this bug was closed (says that the error still shows for dbus programs). per Eric Hammond in http://groups.google.com/group/ec2ubuntu indicates this as a non-issue.
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
Hi there, I m getting a error
Hi there,
I m getting a error while starting the server.
Jun 19 01:18:31 m4manas bcfg2-server[4682]: Failed to read file probed.xmlJun 19 01:18:31 m4manas bcfg2-server[4682]: Loading experimental plugin(s): Packages
Jun 19 01:18:31 m4manas bcfg2-server[4682]: NOTE: Interfaces subject to change
Jun 19 01:18:31 m4manas bcfg2-server[4682]: Failed to bind to socket
Jun 19 01:18:31 m4manas bcfg2-server[4682]: Server startup failed
re: Hi there, I m getting a error
Can you tell me what AMI you launched?
What does "sudo /etc/init.d/bcfg2-server status" say?
Thanks,
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
root@m4manas:~# sudo
root@m4manas:~# sudo /etc/init.d/bcfg2-server statusbcfg2-server (pid 5121 4517 4241) is running...
root@m4manas:~# sudo bcfg2 -vqed
{'profile': False, 'file': False, 'filelog': None, 'dryrun': False, 'verbose': True, 'extra': True, 'ca': '/etc/bcfg2.crt', 'args': [], 'remove': False, 'bundle': [], 'quick': True, 'user': 'root', 'key': '/etc/bcfg2.key', 'setup': '/etc/bcfg2.conf', 'password': '****', 'help': False, 'retries': '3', 'omit-lock-check': False, 'certificate': '/etc/bcfg2.crt', 'paranoid': False, 'indep': False, 'decision': False, 'cache': False, 'serverCN': None, 'servicemode': 'default', 'encoding': 'ascii', 'decision-list': False, 'kevlar': False, 'debug': True, 'server': 'https://localhost:6789', 'drivers': ['Action', 'APT', 'Blast', 'Chkconfig', 'DebInit', 'Encap', 'IPS', 'FreeBSDInit', 'FreeBSDPackage', 'launchd', 'Portage', 'POSIX', 'RPMng', 'RcUpdate', 'SMF', 'SYSV', 'YUMng'], 'interactive': False}
Server failure: Protocol Error: 401 Unauthorized
Failed to download probes from bcfg2
Server Failure
.
Thanx a ton.
re: sudo
It seems that there may be 2 copies of bcfg2-server running. You can check by running:
"ps -ef | grep bcfg2"
If you do see 2 process of bcfg2-server running, I would stop both and start a new instance (kill -9 PID - PID being the process id of bcfg2-server) . There's was a bug in bcfg2/ubuntu where /etc/init.d/bcfg2-server stop or /etc/init.d/bcfg2-server restart wasn't actually stopping the process.
Hope this helps,
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
Thanx a ton. It did worked. I
Thanx a ton.
It did worked. I am sure it would help a lot of people like me.
same error here
I just did this over the weekend and I received the same error. I checked but did not have an additional bcfg2 server running.
I started entire process over, including re-deploying the slice (i'm on linode 2048 Ubuntu 10..04).
In my second try, I accidentally re-ran "sudo bcfg2 -vqed", meaning I ran this once, got the error above about failing to download the probes, and then re-ran it and then it worked fine.
Later in the process, I also ran into another problem, which was the /etc/mercury/mercury.make file was not there. So the drush cmd failed:
sudo drush make --working-copy /etc/mercury/mercury.make /var/www/I looked on my other linode (where I had done this successfully 3 weeks ago) and I also did not see it. I looked in /root home dir and the mercury.make file was in there. I copied to /etc/mercury and re-ran drush and all was well.
I am not sure how the file was not where it needed to be. I ran all cmds as a normal user, putting sudo at the beginning. I ran all of the cmds ot of the normal user's home directory. So not sure how mercury.make got into /root
Otherwise, Mercury is AWESOME! My clients are even noticing how much better things are. Excellent Job!!
divbox
Hi there I was following your
Hi there
I was following your step by step tutorial
http://groups.drupal.org/node/70268
I do not understand about AMI. I m new bee to Mercury.
Thanx
re: Hi there I was following your
Are you using one of our pre-created Mercury AMIs (listed at the top if this thread)? If so, you do not need to follow the instructions at http://groups.drupal.org/node/70268 - those instructions are for creating your own Mercury instance.
You might want to check out http://en.wikipedia.org/wiki/Amazon_Elastic_Compute_Cloud for the basics on Amazon Web Services (AWS), Elastic Cloud Computing (EC2), and Amazon Machine Images (AMI).
Hope this helps,
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
init.sh
Hi guys.
I was following this tutorial (http://groups.drupal.org/node/70268) to setup mercury on my local ubuntu server and got confused at the step 10 - Run /etc/mercury/init.sh.
After completing all the steps, there is no such file on my system. Please help me out here. What have I missed?
re: init.sh
Can you run sudo bcfg2 -vqed again and take note of any errors? Also, please paste the last couple of lines from bcfg2 -vqed including the managed and unmanaged entries.
Thanks!
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
Thank you, Greg. I've tried
Thank you, Greg.
I've tried the new version of instructions (using Hudson), and everything went just perfect. Works perfect as well. However, you should mention about --uri key for mercury_cron project in Hudson for those who use mercury for multiple sites.
Thanks again, and good luck.
Mercury on centos 5.5
Hi Greg,
I just tried merucry1.1 on centos 5.5/rackspace.
I am not able to get bcfg2-server started correctly. Your doc suggest waiting for serve_forever() [start] but this message never came to my syslog.
Also after a long time when i stopped tail -f and checked the status of bcfg2-server using /etc/init.d/bcfg2-server status its giving me this
"bcfg2-server dead but pid file exists"
Its a fresh new server and nothing else was done apart from the mercury installation. Earlier i tried merucry on lucid server but server load was going up in every 3-4 hours. I tried to fix this one but was not able to do it even after tweaking mysql, apache thats why i thought of trying centos.
Let me know if you have any report on this.
Regards
Sreyas
tried on new server
Just now recreated the server tried again.
Jul 23 10:04:33 server bcfg2-server[1210]: Packages: File read failed; falling back to file downloadJul 23 10:04:33 server bcfg2-server[1210]: Packages: Updating http://dl.iuscommunity.org/pub/ius/stable/Redhat/5/x86_64/repodata/filel...
Jul 23 10:04:33 server bcfg2-server[1210]: Packages: Updating http://dl.iuscommunity.org/pub/ius/stable/Redhat/5/x86_64/repodata/prima...
Jul 23 10:04:34 server bcfg2-server[1210]: Packages: File read failed; falling back to file download
Jul 23 10:04:34 server bcfg2-server[1210]: Packages: Updating http://dl.iuscommunity.org/pub/ius/stable/Redhat/5/i386/repodata/filelis...
Jul 23 10:04:34 server bcfg2-server[1210]: Packages: Updating http://dl.iuscommunity.org/pub/ius/stable/Redhat/5/i386/repodata/primary...
Jul 23 10:04:35 server bcfg2-server[1210]: Packages: File read failed; falling back to file download
Jul 23 10:04:35 server bcfg2-server[1210]: Packages: Updating http://yum.fourkitchens.com/pub/centos/5/noarch/repodata/filelists.xml.gz
Jul 23 10:04:35 server bcfg2-server[1210]: Packages: Updating http://yum.fourkitchens.com/pub/centos/5/noarch/repodata/primary.xml.gz
Jul 23 10:04:36 server bcfg2-server[1210]: Failed to read file probed.xml
after this waited for half an hour and still no message was coming. Quit from tail command and check the status of bcfg2-server and it was same again
bcfg2-server dead but pid file exists
Please help
Regards
Sreyas
re: tried on new server
Sreyas,
Something is killing the bcfg2-server - are there any other bcfg2-server related messages in /var/log/messages ("Failed to read file probed.xml" is not a fatal error and is expected before "bcfg2- vqed" is run)?
As for the Lucid CPU issue, it's a know bug with Ubuntu Lucid: https://bugs.launchpad.net/pantheon/+bug/574910
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
I can no longer find these AMI's at Amazon
I have tried to look for these AMI's and can't find them anymore. Did you remove them for some reason?
us-east:
ami-294aa340 pantheon-amis-us-east/mercury-pressflow-drupal-lucid-32-1.1-Beta2.manifest.xml
ami-554aa33c pantheon-amis-us-east/mercury-pressflow-drupal-lucid-64-1.1-Beta2.manifest.xml
us-west:
ami-2b8ddc6e pantheon-amis-us-west/mercury-pressflow-drupal-lucid-32-1.1-Beta2.manifest.xml
ami-298ddc6c pantheon-amis-us-west/mercury-pressflow-drupal-lucid-64-1.1-Beta2.manifest.xml
eu-west:
ami-6d3c1619 pantheon-amis-eu-west/mercury-pressflow-drupal-lucid-32-1.1-Beta2.manifest.xml
ami-673c1613 pantheon-amis-eu-west/mercury-pressflow-drupal-lucid-64-1.1-Beta2.manifest.xml
ap-southeast:
ami-27e39c75 pantheon-amis-ap-southeast/mercury-pressflow-drupal-lucid-32-1.1-Beta2.manifest.xml
ami-25e39c77 pantheon-amis-ap-southeast/mercury-pressflow-drupal-lucid-64-1.1-Beta2.manifest.xml
Symetrik Design
Drupal Consulting
http://www.symetrkdesign.com
re: I can no longer find these AMI's at Amazon
They are still there and still public. You should be able to find them by searching for the AMI id.
Hope this helps,
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
I'm sorry to contradict, but
I'm sorry to contradict, but they are not there.
I searched for specifically the AMI's before I posted originally.
If I search for Pantheon or Mercury, I can only find the Mercury 1.0 32bit and 64bit on US East coast.
If you are seeing them, can you send me the direct link to EU-West 32bit?
Symetrik Design
Drupal Consulting
http://www.symetrkdesign.com
re: I'm sorry to contradict, but
I'm not aware of a url being associated with AMIs.
However, if I go to Amazons EC2 console, select EU-WEST, I can see the following 32-bit AMI:
AMI ID: ami-6d3c1619
Source: pantheon-amis-eu-west/mercury-pressflow-drupal-lucid-32-1.1-Beta2.manifest.xml
Visibility: Public
Hope this helps,
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
Ok, to clear this issue up.
Ok, to clear this issue up. They are there!
I was searching through the AMIs at this URL
http://developer.amazonwebservices.com/connect/kbcategory.jspa?categoryI...
They don't appear through the search mechanism.
HOWEVER,
When I logged into the AWS console, I was able to find them with no problem.
Thanks.
Symetrik Design
Drupal Consulting
http://www.symetrkdesign.com
No other errors were there
Greg
No other errors found on the logs. its just bcfg2 starts and go into the same state. I am again and again trying different rackspace cloud servers. I have removed centos.. now working with karmic..
One more thing I would like to check... What all does mercury profile changes?? I just want to check after upgrading to mercury profile is there any problem if we go back to basic drupal. Because when i had problem with mercury profile switched back to drupal but after that site has been showing multiple issues!!
Is there any chance mercury changed something on my website??
Regards
Sreyas
re: No other errors were there
Looks like you found the solution - i didn't know that bcfg2-1.1.0-rc4 had that issue (i have yet to see the bug appear).
As for Karmic, we haven't configured Mercury 1.1 for Karmic, and likely won't since Lucid is an LTS. Hopefully, the CPU bug will be fixed soon.
The mercury install profile adds some memcached and varnish config the end of the settings.php file and configs pressflow for apache solr, memcached and varnish. Since these are settings that would be overwritten by importing a site (until we finish the import script), the install profile can't change your existing site.
Hope this helps,
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
bdfg2-server not starting resolved.
http://groups.drupal.org/node/83864#comment-261914
What is the proper process for updating
Hello,
I've finally figured out how to get Mercury 1.1 beta up and running at AWS using a 32-bit small instance (ami-6d3c1619).
Some points to note:
1) When I logged into Drupal, I found that the Apache Solr and Varnish modules were out of date. I also ran apt-get update and found that several Ubuntu packages were out of date.
What is the process for updating my AMI? It seems like there is quite a bit going on (Ubuntu, Drush, BCFG, Mercury, etc.)
Should I run ???
A) Drush Updates
$ cd /var/www$ sudo drush up (choose yes to update)
$ cd /var/www/sites/all/modules/apachesolr
$ svn checkout -r22 http://solr-php-client.googlecode.com/svn/trunk/ SolrPhpClient
(When I run drush up, it seemed to take care of varnish errors that appear on admin pages, and admin/status page is all GREEN after the Drush updates.)
B) Run ubuntu updates
$ sudo apt-get update(This seems to download alot of packages/files)
$ sudo apt-get upgrade(This upgrades several packages, but after it is finished my Apache Solr server no longer works. Further investigation appears to show that tomcat6 cannot start)
C) Run Mercury update script (as mentioned in the directions in the top of this document)
$ sudo /usr/local/bin/update_mercury.sh(This doesn't seem to do much)
D) Run bzr updates (as mentioned in /etc/mercury/init.sh)
cd /var/www/sudo bzr merge --force
cd /var/lib/bcfg2sudo bzr merge --force
Any comment on this would be appreciated.
I'm just trying to update anything that is out of date since the AMI was packaged, and learn how I should be running long term preventative maintenance on the instance. If I've got it all wrong or barking up the wrong tree, I'd love to learn how to do it correctly.
Symetrik Design
Drupal Consulting
http://www.symetrkdesign.com
re: What is the proper process for updating
a) Drush Updates - there is no mechanism in Pantheon (Mercury) to do this - so this is the best way to update Drupal modules.
b) Run ubuntu updates - This should be is handled by the update_mercury script and it does it in a way that doesn't overwrite pantheon-applied config files (such as those used by tomcat).
c) Run Mercury update script - This script also downloads Pantheon (mercury) config file updates to applies them using bcfg2. A logfile is created in /root/update_mercury.log
d) Run bzr updates - the update_mercury script handles bzr updates for the /var/lib/bcfg2 directory, and the update_pressflow script handles bzr updates for /var/www.
I'm not sure why you're getting errors in updating the /var/www directory - I don't see that error when running update_pressflow on a fresh 1.1 beta AMI.
The error from trying to update the /var/lib/.bcfg2 directory probably comes from some housekeeping I did a while back that re-created some of the branches. This is what I did to fix the problem:
cd /var/lib/bcfg2; sudo bzr pull --overwrite lp:pantheon/1.1/If you get errors about Packages.moved and Probes.moved, you can:
sudo rm -rf Packages.moved Probes.movedsudo bzr resolve Packages.moved
sudo bzr resolve Probes.moved
And then run /usr/local/bin/update_mercury to apply any updated configs. BZR will now use the correct launchpad url.
Hope this helps,
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
Updating Pantheon/Mercury and Ubuntu
Greg,
1) Thanks for these updating instructions. I found them by chance, which is a pity given how important they are. Therefore I'd like to ask you where can we find the most current documentation about Pantheon/Mercury. It used to be at getpantheon.com but those links (I had some bookmarked) don't work anymore. There is some documentation at http://groups.drupal.org/pantheon/documentation but that is incremental (much of it is not relevant anymore).
2) I am afraid to ask such a basic question, but what is the difference between "Pantheon" and "Mercury"? I gather Pantheon shall include more Drupal gods in the future, but for the time being Mercury == Pantheon?
I never know whether to talk about Pantheon or Mercury when referring to the techniques you guys develop for performant cloud installs of Drupal. I think I can see other people being a bit confused, and even you, in your above post, painstakingly add "Mercury" after "Pantheon" to make things clear.
If this is too basic a question, I apologize, but please provide an URL where I and others can read up more.
Thanks,
Tomáš (vacilando)
---
Tomáš J. Fülöpp
http://twitter.com/vacilandois
re: Updating Pantheon/Mercury and Ubuntu
Currently, http://groups.drupal.org/pantheon/documentation is the best place for documentation. We'll be adding much more to getpantheon.com soon - in fact it's my next task. I'll also add the above to g.d.o/pantheon/documentation. If you find anything else useful that belongs there, let me know.
There is no such thing as a "too basic" question (in my opinion) and others have been confused too. At one time, there was a distinction between the 2 names but that was confusing so now we just call is Pantheon.
Hope this helps,
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
Varnish Errors on SIte
HI Greg, everyone.
After spinning up the AMI on EC2 and then going through the site DB setup process using the root account with no pw I get this errors when viewing the site:
* Notice: Use of undefined constant VARNISH_NO_CLEAR - assumed 'VARNISH_NO_CLEAR' in include_once() (line 4 of /var/www/sites/all/modules/varnish/varnish.module).* Notice: Use of undefined constant VARNISH_DEFAULT_CLEAR - assumed 'VARNISH_DEFAULT_CLEAR' in include_once() (line 5 of /var/www/sites/all/modules/varnish/varnish.module).
* Notice: Use of undefined constant VARNISH_SELECTIVE_CLEAR - assumed 'VARNISH_SELECTIVE_CLEAR' in include_once() (line 6 of /var/www/sites/all/modules/varnish/varnish.module).
I've not found anything here that talks about this so can't work out what's up. Any suggestions please?
Sounds like a bug in Varnish
Sounds like a bug in Varnish module
http://drupal.org/project/varnish
Best to search for it there
--
Dave Hansen-Lange
Director of Technical Strategy, Advomatic.com
Pronouns: he/him/his
Well they appear to be farily
Well they appear to be farily low level warnings dalin, but thanks. Will check.
So I'm not the only one here
So I'm not the only one here whose php error reporting started giving Notice level warnings last month? I think it was just after the 6.19 update, or maybe it was a module that did it. I'm pretty sure I didn't touch php. I started getting all sorts of Notice level warnings in my Drupal log about a variety of modules and one core file. Modules that I hadn't updated in a long while.
I had to change my php.ini from
error_reporting = E_ALL & ~E_DEPRECATED
to
error_reporting = E_ALL & ~E_NOTICE
in order to get rid of them. I still have no idea what started em.
Can't find anything about it on d.o. No doubt limited impact because it's related to PHP 5.3.x.
Are you using PHP 5.3.2 also, ClickInfluence?
5.2.10-2
Hi Brian, here's what I'm using. It's on the default 1.1-beta AWS image for Mercury
$ php --versionPHP 5.2.10-2ubuntu6 with Suhosin-Patch 0.9.7 (cli) (built: Oct 23 2009 16:30:10)
Copyright (c) 1997-2009 The PHP Group
Zend Engine v2.2.0, Copyright (c) 1998-2009 Zend Technologies
I imagine it's the default setting for PHP on ubuntu..
thanks for the php.ini tip!
PHP 5.3 - my 2 cents.
PHP 5.3 seems to be almost over the line as far as Drupal is concerned.
I find that using dev versions of most modules helps.
That said, most of my client sites are on PHP 5.2x.
Though it is a bit of a pain, overall I'd say that 5.3 is better then some dodgy 5.2 downgrade.
Benchmarking
Hi Greg,
Any benchmarking results to share for this AMI. Please share.
Thanks
Xavier
re: Benchmarking
Xavier,
Sorry, we've moved on to getting Pantheon 1.2 ready for release. I hope to have some benchmarks for 1.2 around the time we do release.
Greg
--
Greg Coit
Systems Administrator
http://www.chapterthree.com
Availability zones
Looking forward to the new Mercury 1.2 release, when they are available.
--
G