Switched to Serverless Aurora from a cluster over the weekend for availability and scaling.
Ran fine for two days. Then for two 10 min periods in the last two days, we get "MySQL server has gone away" errors on about 30% of queries.
Have been running same code on Aurora clusters for two years with no issues.
No scaling was occurring, there were no RDS events or log entries, no change in metrics during the period.
The queries are normally sub-second, and the error returned much earlier then client timeout.
There is ongoing traffic (so not a warmup issue) and multiple EC2s accessing DB at the time (so not related to a specific client).
Some other posts in the last month indicate similar problems; has anyone else seen this?
Any ideas on the cause or a fix?
Edited by: Vueocity on Mar 4, 2020 9:33 AM
Edited by: Vueocity on Mar 4, 2020 9:35 AM
