Posted by dmartens on August 3, 2010 at 8:01pm
Thanks to the developers of these capabilities for all the wonderful work they've been doing.
As someone who is currently experimenting with Pantheon/Mercury, and who also has an interest in Aegir, I wonder whether the rumor I read about a forthcoming version of P/M with Aegir integrated is true. That would be great, of course, and may affect how I approach my current efforts.
Without sounding demanding, can anyone advise whether this rumor is in P/M's current plans and if those plans involve a release "soon" versus "someday" ? FWIW, I did look for a timeline of anticipated features for Pantheon, although I may not have looked in the right place.
thanks
Dave
Comments
It's more of a "someday"
Now that the Alpha9 is out and solid, we're thinking about this more. However, the current focus of Pantheon development is stability of the current platform. Adding an "Aegir Installed" version will have to wait until after that, so it won't happen until the Fall at soonest.
However, there's been a lot of demand for this from developers. I would love to have a rock-solid Pantheon+Aegir 0.4 (final release) before the end of the year, but can't promise anything at this point.
https://pantheon.io | http://www.chapterthree.com | https://www.outlandishjosh.com
One of those interested in Pantheon+Aegir :-)
Just adding my voice to those interested in seeing a clean pre-packaged Pantheon+Aegir combo. But till then(Fall's not that far away ...) When I get a chance, I'll probably try some experimenting and see I can find all the stumbling blocks like overlapping cache and see I can come up with something that doesn't blowup in my face.
Either way, thanks for the work, I can't wait till I need that power :-)
I'm strongly tempted to add
I'm strongly tempted to add "some" Mercury stuff to my all-in-one installer for Aegir on Nginx (Lucid, Karmic and Lenny compatible). Before CPH DrupalCon :)
Or maybe go on top of Mercury
Or maybe go on top of Mercury instead.
This way you get to take advantage of all their work. Although I think there are a few twists that have to be taken care of with the cache for APC & memcache and an ID that has to be set for Solr to keep things unique.
I can't use Mercury as a base
I can't use Mercury as a base system because I have different goals in my project. I don't use Apache and Varnish and don't intend to copy the Mercury idea, which is completely different (since I don't plan to create anything like ready to deploy AMI etc.) My project is about giving ready to use hosting system based on Aegir and Nginx, but done the old way - by running shell installer, so it is not about highly scalable systems like Varnish based. The idea is however, to create a "turnkey" solution, easy to install on any VPS with supported OS with performance similar to Mercury, but not scalable as Mercury (since I don't have now any plans to use Varnish).
Thanks for the reply. What
Thanks for the reply. What you say makes sense. We each have diff goals/requirements for our projects. But that's also part of where the power of open-source comes in. Even with diff end results, we can share and re-use the knowledge. Good luck and thanks for all your informative responses in the forums.
Thanks
Makes perfect sense - your response was helpful.
Resource
I created a wiki to track tips on setting up Aegir + Pantheon.
I know that a few people currently have it up and running. You just need to install Aegir yourself on top of Pantheon.
http://groups.drupal.org/node/55033
Ive been running them together for some time now.
I run alpha7 in mercury 1.0 and have run Aegir in it for many un-buggy perfect happy months... :)
I was unsuccessful at getting alpha8 to work with either mercury 1.0 or 1.1. However I was able to upgrade my alpha7 to alpha8 (on my server) with no problems. A fresh install on a fresh mercury wouldn't work.. killed I mean completely killed mysql db after 2nd verify of the first main Aegir site. I didnt have too much time to muck with it or figure the prob.. it was a clients server... so I just left Aegir out of the mix and went back to the boring standard way :(
Funny thing is it works great on my server which is an identical Mercury 1.0 AMI .. only difference is it was an upgrade and not a fresh install .. ive had Aegir on that box since alpha3 ... I will try the new alpha on a fresh install after they fix the drush issues though.
Hopefully it was a one off thing (although I tried 6 times) cause I love me some Aegir in Mercury.
I asked around on irc and got some pretty good tips .. the biggest being to use Debian which I scoffed at but now im not so sure. Ubuntu bit me about a week ago when I was setting up OpenSim .. it just wouldn't work ... so I decided to try Debian and it worked instantly with no fuss.
I may do a manual install of Mercury on Debian to see how it feels .. so far it feels the same but better somehow... in Debian I hear they have a better support/update setup than ubuntu though IDK really; I always thought they were nearly identical.. until recently.
I wonder if there was a specific reason Mercury folks chose Ubuntu over Debian?
Cheers!
Joshua
or "someday" aegir will be running fast out of box :)
Aegir is getting performance and scalability features, look at overview of 0.4 Alpha 9 http://bit.ly/9SUvWA
they know about alpha 9
the sponsored the release =)
I've noticed
just wanted to mention another scenario per original question ;)
It is fast already IMO
The power Mercury can really wield is in the platforms and not the actual Aegir app which performs fast enough IMO though more speed is always nice ;)
With full integration one could imagine automatic provisioning of solr & memcached cores, I think I read recently someones doing just that already.
Or even more exciting tie into the killer AWS api and provision whole servers, HD's etc for a site if ya want all in the same app. Makes me drool! lol love Aegir!
Solr support is coming
I'm working on automagic provisioning of Solr cores. The Solr side of stuff is done and work, you can grab the code from my blog - http://davehall.com.au/blog/dave/2010/06/26/multi-core-apache-solr-ubunt... I just need to work on the aegir side of things. Two things are slowing this down, paid work commitments and the constant changes to the aegir code base. I hope to have all the Solr support stuff in git by the time I arrive in Copenhagen.
yes it was your blog I was
yes it was your blog I was reading .. very cool!!
Cheers!
Varnish Barrier
I handled the memcache and solr unique configs for each of my sites by placing conditionals in /home/aegir/config/includes/global.inc . The Varnish conditionals I attempted using Varnish Cache Language scripts. Without the VCL scripts a request to any of my sites with the same URI would sporadically serve up pages from another of the sites on the server due to the cache ambiguity of the request. Writing these scripts was hellish for me and I abandoned the experiment altogether. If someone has a solution to multiple vhosts + Varnish I would be very grateful.
Hm, I didn't noticed any
Hm, I didn't noticed any Varnish Barrier when running Aegir from under Mercury, but I had similar problems with non unique memcache_key_prefix, and works after following hack:
Am I lost anything?