When GitLab starts up, the API isn't immediately available, even though the service has started. If label caching is enabled, API connection failures should result in an exponential backoff sequence capped at some amount after which the service should simply fail.
Optionally, status information about the backoff should be posted to the default channel for the Slack webhook so that there's some notification of this happening.
When GitLab starts up, the API isn't immediately available, even though the service has started. If label caching is enabled, API connection failures should result in an exponential backoff sequence capped at some amount after which the service should simply fail.
Optionally, status information about the backoff should be posted to the default channel for the Slack webhook so that there's some notification of this happening.