Fix proxysql-admin zombie processes issue
Description
Smart Checklist
Activity
KennT January 26, 2021 at 10:04 AM
Added a new flag --use-stdin-for-credentials.
If this flag is not seet, then process substitution will be used to pass parameters to the mysql client. (this is the default)
If this flag is set, then stdin will be used to pass parameters to the mysql client.
Slava Sarzhan December 10, 2020 at 7:43 AM
it is strange why it appeared after 5660a6b653aa3f2ca0c1d4292531a60ba22afac8 commit.
KennT December 10, 2020 at 5:18 AM
That information is enough. It looks like the processes are zombie processes, they've been killed and are waiting for the parent. It looks like one is being created for each time we do process substitution. I'm looking into this to see what we can do.
https://askubuntu.com/questions/201303/what-is-a-defunct-process-and-why-doesnt-it-get-killed
Slava Sarzhan December 9, 2020 at 4:53 PM
I have the output of the following command
which I collected for . Is it ok for you ? Or I need to reproduce the issue once again and collect the output of "ps aux"
.
KennT December 9, 2020 at 11:33 AM
Can you also supply the list of processes? Maybe just adding a "ps aux" at the end of your script.
Hi,
In cloud team we have the following bug report https://jira.percona.com/browse/K8SPXC-541 . I have investigated this issue and can confirm that we have a lot of proxysql-admin zombie processes in proxysql docker in case proxysql ver >= 2.0.13. The root of the issue is the following contribution https://github.com/percona/proxysql-admin-tool/commit/7f08e5c7e35fa6eb0b6b897b98969606e2c330b5 . I have rolled back this commit and the issue disappeared.
Could you please check it and fix.
Thank you.