"MySQL server has gone away" 3 nights in a row, across two SEPARATE projects

[Robin Donaldson]

We have a simple, normal App Engine (PHP) with PDO connection to our Cloud SQL DB.  We have a Staging project (Cloud SQL 2nd gen) and Production project (Cloud SQL 1st gen).

We get stale connections from time to time, so we check whether the connection is alive at the start of a request, and get a new connection otherwise.  This code works and we see this retry succeeding in Staging and Production all the time.

However, for the past three nights at almost the same time, SYNCHRONIZED between our two separate Staging and Production projects, we now get all our connections from GAE to Cloud SQL are stale.  This lasts for around 10 minutes.  It starts at the same time and ends at the same time for our two separate projects.  Tonight I restarted the database twice during this outage and it DIDN'T FIX IT.  

I conclude this is an outage on the part of Google Cloud because it's synchronized between projects, because it's happening every night, and because restarting the DB (which refreshes the connection pool) did NOT fix it.  

For these reasons I need to talk with someone at Google Cloud immediately.  We have a reputation to keep as a young business and we are getting customer complaints about lack of service.  As CTO I am left in the position where I do not know what to do.  It seems that this issue is completely outwith my ability to fix now.

Robin Donaldson

[Taher (Cloud Platform Support)]

Hi Robin,

Please check the feedback given in your other post in this group.