Aborted connection 47625 to db | Read Replica

[abhishek singh]

For my CloudSQL DB, Read-Replica of PROD, i am getting frequent aborted connections error. We use Read-Replica for reporting purpose, We use SQL proxy to connect from MS-SQL server ( hosted in Azure) along with C-data in between for fetching data from CloudSQL to MS-SQL server. 

SQL quires executes for 30-180 mins and are running for last 1 year. In last 4-5 days , our reporting SQL's start hanging after running for 20-30 mins and did not respond. 

Read-Replica logs show are is a increase in aborted connect error.

Logs:

====

{

 insertId: "xxxxxxxx"  

 logName: "projects/xxx-xxx-db/logs/cloudsql.googleapis.com%2Fmysql.err"  

 receiveTimestamp: "2020-03-30T13:07:53.626203743Z"  

 resource: {…}  

 severity: "INFO"  

 textPayload: "2020-03-30T13:07:52.901100Z 47625 [Note] Aborted connection 47625 to db: 'XXXXDB' user: 'readonlyusr' host: 'cloudsqlproxy~xx.xx.xx.xxx' (Got an error writing communication packets)"  

 timestamp: "2020-03-30T13:07:52.901563Z"  

}

[Katayoon (Cloud Platform Support)]

Hello Abhishek,

There was an infrastructure issue reported on March 30, 2020 and on March 26, 2020 which might affect your Cloud SQL instance. However, "Aborted connection nnnn to db:" error usually indicates that your application is not terminating connections properly. If the issue is still occurring, you may take a look at the best practices for connection management; and If you have already taken the best practices and it hasn't alleviated the issue, you may report it to the Google Cloud Platform Issue Tracker so that we would be able to further investigate it.

Note that GCP groups are reserved for general discussions on Google Cloud Platform-end products. For reporting issues and feature requests, you should create a ticket via the GCP Issue Tracker.