[MDEV-31593] Constant aborted_clients due to read/write errors Created: 2023-06-30  Updated: 2023-10-24

Status: Open
Project: MariaDB Server
Component/s: None
Affects Version/s: 10.6.12
Fix Version/s: None

Type: Bug Priority: Major
Reporter: zddshp63 Assignee: Unassigned
Resolution: Unresolved Votes: 0
Labels: None
Environment:

Ubuntu 22



 Description   

[Warning] Aborted connection 253668 to db: 'xxx' user: 'xxx' host: 'localhost' (Got an error writing communication packets)
[Warning] Aborted connection 255472 to db: 'xxx' user: 'xxx' host: 'localhost' (Got an error reading communication packets)

I have done everything everything possible to resolve the issue including.

Raising max allowed packets
Raising php timeouts
Raising mysql timeouts
Disabling all wordpress plugins
Reinstalling wordpress / checksumming wordpress
Disabling redis
Migrating to new server to ensure no hardware issues

Mariadb is running locally to wordpress so there are no network issues. I can see the aborted_client count increasing every second in sync with these read/write errors. The error logs at max verbosity do not give enough information. I've looked at php error logs, open lite speed error logs, wordpress error logs, performance schema. No information is given as to why connections are being aborted.

When I look at mariadb general access logs I can see the last query before the connection is aborted is always a simple select query.

I have had several mariadb database admins look at the issue and everyones conclusion is that it is a bug.


Generated at Thu Feb 08 10:25:02 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.