MySQL server has gone away about delayed_job

[stephankaag]

Same issue overhere.

[scottj97]

See line 64 of lib/delayed/command.rb. Looks like it's trying to reconnect.

When exactly do you have a problem? I tried restarting my app after launching delayed_job, and that did not seem to cause any problems. Delayed jobs still get run.

[sheldonh]

When the MySQL server times the connection out, as per the MySQL server's configured wait_timeout setting.

[scottj97]

Doesn't DJ continuously poll the MySQL server? Why would it time out?

[sheldonh]

Aha! Found it. The application uses two databases (on two different servers). DJ isn't being timed out by the server on which the delayed_jobs table exists; but one of the delayed jobs accesses a legacy database, on another server, and that's where the timeout occurs.

[bkeepers]

Sounds like this isn't necessarily a dj issue. Reopen if I'm wrong

[andrewroth]

what's the workaround?

[bracken]

I had this problem too because our delayed_job table is in a different database than our app. For those looking, here is my simple fix: http://github.com/bracken/delayed_job/commit/b73a7e561b89fc09ae77bb9c1cfc680062867e19
You could do the same thing in your custom daemons so that you can still use the main delayed_job branch.

[bkeepers]

For those still having this issue, make sure you try this first: http://guides.rubyonrails.org/2_3_release_notes.html#reconnecting-mysql-connections

[sheldonh]

Bracken's take on the issue makes the world of sense. Our problem is also from using multiple databases, where the jobs aren't stored in the database that the in-job code interacts with.