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
Problem: Graceful recovery tests were failing in the pipeline
Solutions:
- Increase timeout when testing for passing/failing traffic after a container or pod restart.
The timeout was set to 20 seconds, with a polling interval of 500ms, but the per-request
timeout was 10 seconds. This means we were only retrying the request twice when a request
timed out because it couldn't connect. Increasing the overall timeout to 60 seconds increased
the number of retries and eliminated the errors.
- Remove the kubectl drain flag --delete-local-data since it no longer exists and
replace it with --delete-emptydir-data.
0 commit comments