// the find
mhfs/sidekiq-failures
Keep track of Sidekiq failed jobs
sidekiq-failures records failed Sidekiq jobs and adds a Failures tab to the Web UI, so you can see what broke without digging through logs. It suits teams already on Sidekiq who want Resque-style failure browsing and don't want a separate error tracker for background work.
The tracking mode is set per worker to :all, :exhausted, or :off, with a global default in the server config. That matters because a job with 25 retries under :all writes up to 25 entries, while :exhausted keeps only the failures you would actually act on. The 1000-entry cap is on by default and can be raised or disabled explicitly, so the list has a bounded footprint unless you choose otherwise. It uses Sidekiq's middleware chain and custom web tabs, which are public extension points, so there is less to break across Sidekiq upgrades.
The last push was August 2025, more than a year before today, so check the open issues before adopting it on a current Sidekiq. Eviction is count-based with no retention by age, so one noisy worker can push out every other failure in the list. The programmatic surface is thin: you can count, clear, and reset the stats, but there is no lookup or filtering. The dependency is a floor (Sidekiq >= 4.0.0) with no stated upper bound, and the README doesn't say which Sidekiq versions CI covers.