Postgresql Flexible Server throwing error 08P01

Question

I'm running a Ruby on Rails application against Postgres servers. For a long time, I've struggled with delayed_job falling over. The error messages in the delayed_job stack traces are ambiguous, but they seem to suggest that the database instance is being interrupted. The other morning, delayed_job and three other, long-running housekeeping jobs were all killed. Thanks to other issues, I've started logging on my production instance, and it looks like these process-killing problems are happening when the database logs an error code of "08P01". According to PSQL docs, that's a connection exception with a protocol error. Nothing else should be connecting to the database right now, while the "housekeeping" is going on. I'd cast suspicion on the backups, but they happen every day, and delayed_job runs for "many" days until someone complains that they didn't get email from the system, and I restart it. This feels like some sort of housekeeping that Microsoft is doing on Azure, restarting the database, but I don't know what that would be, let alone how to stop it (or at least schedule it). Does anyone know what could be interrupting my database connections every "several" days?

Answer 1

Hi, @David Krider Welcome to the Microsoft Q&A platform, thanks for posting the question and for using Azure Services.
In the Flexible server, you do have the possibility to schedule 1hr of maintenance window with a custom day. We also guarantee that this does not happen more than once a month.
For long-standing troubleshooting sessions, I would recommend filing a support case.

Please let me know f you need any help raising a support case or any additional questions.

Regards
Geetha