Closed resworld closed 2 months ago
Hi @resworld .
If connections and COM_CHANGE_USER
all started timeout, this issue is likely related to a backend issue.
The app has a downtime because proxysql shunned the backend, and proxysql shunned the backend because the backend was timing out both the creation of new connections and COM_CHANGE_USER
Hi @renecannao and thank you for the fast reply,
We don't see anything in the backend logs in that certain time. We will dig deeper in the galera and try to find the root of this issue. Thank you once again!
Hi team,
We have faced an issue with the proxysql that caused a downtime for our site. It seems that the proxysql marked our writer server as shunned and all the queries has been rejected that way. We have noticed that, just before that to happened, it has "timeout error" during COM_CHANGE_USER. We have migrated to proxysql 2.6.3 version а week ago and this error is something new for us. The application is using a lot of different MySQL users (more than 10k) that are still with
mysql_native_password
method for authentication and our proxysql's default plugin is set tomysql_native_password
as well. Problem appear just once, a day after the migration, and after that it doesn't happen again. Can you, please, help us to understand what it could be happened and how can we prevent it from happening in the future?We are using proxysql admin instances deployed in AWS and proxysql satellites running inside the EKS cluster. Both are using same proxysql version:
OS version:
Steps to reproduce:
Logs from the Satellite:
Please let us know if something as information would be needed! We will like to understand what could be configured wrong or how to prevent from happening in future.
Thank you!