Sidekiq: Dead job queue

[mperham]

Sidekiq 3.0 will introduce a dead job queue for jobs which keep failing after the 25th retry.

The DJQ cannot be a memory leak though; it has to be limited somehow. What should the retain policy be? 6 months or 10,000 jobs?

If I configure a job to have 10 retries, should it go to the DJQ on the 11th? I assume yes. If I configure a job to have no retries, should it go to the DJQ upon failure? I assume no.

[mikegee]

How much work would it be to expose these choices as configuration at the application level or in the worker class? Any of the options you have here would be fine as default values.

If I configure a job to have no retries, should it go to the DJQ upon failure? I assume no.

I think this one in particular could go either way.

[mikegee]

I don't see a condition for jobs with no retries. Does that mean you changed your mind and they go into the morgue, too? I just want to make sure the code matches your description of the feature here.

Thanks!

[mperham]

Yeah I forgot to define those rules, also need to doc the feature in wiki.

On Mar 1, 2014, at 4:34, Michael Gee [email protected] wrote:

I don't see a condition for jobs with no retries. Does that mean you changed your mind and they go into the morgue, too? I just want to make sure the code matches your description of the feature here.

Thanks!

—
Reply to this email directly or view it on GitHub.

[heaven]

@mperham Hi, how do we put the job to the morgue if retries are disabled? We're processing payments and if a job fails we don't need it to retry automatically but want to see all failed jobs in the morgue instead. Tried retry: false, retry: 0, and dead: true, but the job doesn't appear in the morgue anyway. Sidekiq 4.2.10.

[mperham]

retry: 0 is how you get a job to skip retries and go straight to DJQ.

[kapildewade]

@mperham Hi, is there a way to customize DJQ size?