stats timeout in 1 hour but actual processing has no timeout

[betelgeuse]

betelgeuse@pena /mnt/checkouts/sidekiq $ git grep EXPIRY | cat 
lib/sidekiq/processor.rb:          conn.setex("worker:#{self}:started", EXPIRY, Time.now.to_s)
lib/sidekiq/processor.rb:          conn.setex("worker:#{self}", EXPIRY, Sidekiq.dump_json(hash))
lib/sidekiq/util.rb:    EXPIRY = 60 * 60

This means that long running jobs will not show up in stats after one hour. @mperham is still some kind of a leftover or what is the logic here?

[mperham]

What is a better expiry time?

[betelgeuse]

I would change it to have no expiry and use del when it's done processing. If the worker crashes there is already cleaning code based on pid. Sounds ok?

[mperham]

I don't like putting data in redis with no expiry. A one week expiry is ok if it is removed when complete under normal circumstances.