This repository has been archived by the owner on Oct 1, 2020. It is now read-only.

Back off when a listener is dead #32

Closed

mezis opened this issue Sep 2, 2016 · 1 comment

Contributor

mezis commented Sep 2, 2016 •

edited

When delivery fails (on whichever error, e.g. 500 from the subscriber), we instantly retry (when the batching deadline has been reached).

Implement a backoff mechanism so that retries happen after 100ms, 200ms, 400ms, 800ms, 1600ms, 3200ms, 6400ms (no higher).

Not a bug per se (everything still works) but this keeps the watch process busy retrying, at 100% CPU. It does still service oether subscribers.

mezis added the bug label

mezis mentioned this issue

Concurrent delivery #38

Closed

Contributor Author

mezis commented Dec 6, 2016

Moved to https://deliveroo.atlassian.net/browse/IP-424

mezis closed this as completed

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.