Display Release Evidence collected_at for Sidekiq reliability
Problem to solve
Sidekiq Reliability
The Sidekiq timing issues are very valid and are important for any scheduling of future jobs. The first potential problem is: the Job fails completely due to a system outage, and the job request is lost from Redis. This should be a low chance of this happening, as while Redis is a in-memory DB, a snapshot is written to disk and restored after a crash.
The second problem is it is not possible to control when a Sidekiq job will run. Even when the job is queued immediately, it is possible that something changes in the Evidence before the job was run, eg: a new Issue is added.
Making jobs 100% reliable is a much larger problem that will likely require tracking calls to Sidekiq and if they succeeded inside Postgres or use Sidekiq Pro. It's quite possible we don't really need this level of reliability. Do we have future jobs (that cannot fail) scheduled elsewhere in GitLab?
Intended users
Further details
Proposal
Suggested approach: display Evidence collected_at
Sidekiq is mostly reliable, and we should assume that the job will run. One approach to handle the Evidence collection time lag is to display the Evidence created_at
(named as Evidence collected at
) on the page next to the Evidence link.
This issue could also be split out for wider discussion.