You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The framework needs a service for identifying liveliness of the GitHub API.
It can be integrated with rate limiting mechanisms.
The idea is to stop/prevent doing "useless" requests and "going silent"
in case of GitHub API degradetion to be "good citizens" and reduce load
on them.
Once it is detected that GitHub API is recovered it's a good idea to
increate client-side throttling gradually so that we don't DDoS them and
also it will give us some time to check whether the liveliness is not
flaky.
I'd start "recovery probes" with Events API requests to identify all events
we "lost" and didn't process, cache those and start going through the
event backlog to get back on track.
N.B. We need to investigate processing data from Migrations API and figure
out whether it's possible to extract missing data for matching events with
webhooks from those.
Also, it's nice to probe GitHub API at startup to indicate whether it's
online in the beginning on the console/log output and make sure that we
can actually start with it being inaccessible.
Oh, and we need to think about graceful shutdown and dumping events which
we weren't able to process to some persistent storage.
The text was updated successfully, but these errors were encountered: