Frontend missing after upgrade [#2102455]

I upgraded from 6.x-1.10 to 6.x-2.0-rc4. There were a few errors along the way, but I think they got sorted out. The only problem (well, it's kind of a HUGE problem) is that my frontend (aegir.example.com) doesn't exist! If I try to go to aegir.example.com, it rewrites to 0.0.0.1.

There's literally no node for the aegir.example.com site in the database. There's a vhost file for it, but I don't think it's been generated correctly. The site files DO exist at /var/aegir/hostmaster-6.x-2.0-rc4/sites/aegir.example.com

My only thought for how to recover from this is to re-import the the site using a command like drush provision-save @aegir.example.com --context_type=site. However, this results in the following errors:

PHP Fatal error:  Call to a member function get_services() on a non-object in /usr/share/drush/commands/provision/Provision/Context.php on line 293

Fatal error: Call to a member function get_services() on a non-object in /usr/share/drush/commands/provision/Provision/Context.php on line 293
Drush command terminated abnormally due to an unrecoverable error.                                                                                             [error]
Error: Call to a member function get_services() on a non-object in /usr/share/drush/commands/provision/Provision/Context.php, line 293

I'm totally stumped. I don't know how to recover from this except to completely purge my Aegir installation, reinstall, and the re-import all of my platforms and sites one by one (uggghh).

Comments

Comment #1

ergonlogic

he/him

English

Montréal, Québec 🇨🇦

CreditAttribution: ergonlogic commented 1 October 2013 at 21:19

Are you using nginx? If so, #1552430: Hostmaster upgrade/migrate doesn't update @server_master drush alias and web server config files might be related.

Comment #2

Dane Powell CreditAttribution: Dane Powell commented 1 October 2013 at 21:36

I'm running Apache, sorry.

Hopefully we can figure out the root cause, but more important to me at the moment is just getting the site back online. If you have any tips for that, I'd really appreciate it.

Comment #3

Dane Powell CreditAttribution: Dane Powell commented 2 October 2013 at 17:52

Well this is bizarre. I went back to troubleshoot more today, and the site is up! I'm not sure what happened overnight to 'fix' it. Must have been some sort of cron task.

However, the frontend isn't actually listed as a site in Aegir. If I try to verify the hostmaster platform, it fails with "Could not find provision alias named: @".

Comment #4

Dane Powell CreditAttribution: Dane Powell commented 2 October 2013 at 18:26

Status:

Active

» Fixed

Looks like the hosting_platform table was corrupt- there was no web server associated with the hostmaster platform.

Everything seems to be working now for me. If you want to leave this open to try to determine the root cause, feel free. Otherwise, I guess this is 'fixed' for me.

My best guess at this point is that the aegir frontend site had actually gone missing somehow prior to the upgrade, and that broke the whole upgrade.

Comment #5

Dane Powell CreditAttribution: Dane Powell commented 2 October 2013 at 21:24

Status:

Fixed

» Active

Well... I just tried to upgrade a few packages using the following command...

drush @hostmaster hostmaster-migrate aegir.example.com /var/aegir/hostmaster-6.x-2.0-rc4-2

and now I can't access my frontend any more- it's redirecting to 0.0.0.1 again!

I tried restarting apache, running drush cc drush, even completely rebooting, with no effect.

Comment #6

Dane Powell CreditAttribution: Dane Powell commented 2 October 2013 at 21:31

If I run drush @aegir.example.com provision-verify, it can't bootstrap the database. Here's a clue: the drush configuration and Drupal root point to the OLD hostmaster platform.

Comment #7

anarcat CreditAttribution: anarcat commented 20 November 2013 at 20:02

Issue summary:	View changes
Status:	Active	» Fixed

this really looks like #2118061: redirections explode after upgrade, try the patch there and verify the hostmaster site, or edit and save the hostmaster node after fixing the vhost by hand or wait for the next 2.x release.