We've been seeing these warnings for quite some time but haven't been able to fully understand why they occur. This was first observed in T310066: Production error: sql-backup failed to start due to no database available but seems unrelated to that original problem.
https://cloudlogging.app.goo.gl/T9nXutUHCrmS5Aje7
The original suspicion was that this could be related to the secondary SQL not having enough memory but seeing that this has now plateaued this is probably not the case.
One alternative thread to investigate could be the errors of dropped connections we are also seeing on the konnectivity-agent https://cloudlogging.app.goo.gl/9jeWzRaxbSaDM7sm6, however when trying to correlate these two errors it seems the konnectivity-agent is actually referring to some other internal service most of the time.
AC
- Figure out whats wrong and document it here
- (Optional) Fix the problem