[MDEV-16924] 10.3.8 FLUSH TABLES WITH READ LOCK hangs indefinitely Created: 2018-08-08  Updated: 2018-08-12  Resolved: 2018-08-12

Status: Closed
Project: MariaDB Server
Component/s: Replication
Affects Version/s: 10.3.8
Fix Version/s: N/A

Type: Bug Priority: Major
Reporter: Sander Pilon Assignee: Unassigned
Resolution: Duplicate Votes: 0
Labels: crash

Issue Links:
Duplicate
is duplicated by MDEV-16923 10.3.8 crashes Open

 Description   

I have a server with 8 parralell slave threads, mode optimistic.
When I start the server, and issue a FLUSH TABLES WITH READ LOCK - that command hangs and eventually crashes the server.

When I change the slave threads to 1 and mode none, the command proceeds.



 Comments   
Comment by Daniel Black [ 2018-08-09 ]

What is the database master/slave version?

Is this occurring on a master or slave?

Do you have a stack trace?
or https://mariadb.com/kb/en/library/how-to-produce-a-full-stack-trace-for-mysqld/ at the time it is hanging.

What is the master/slave config? What form of replication is happening (multisource, gtid, file/pos)?

Comment by Sander Pilon [ 2018-08-09 ]

It is a master-master config with one getting "active" updates. Old style replication file/pos.

This is happening on the master that is replicating almost nothing.

Comment by Daniel Black [ 2018-08-09 ]

Assuming same config/version as MDEV-16923

Comment by Sander Pilon [ 2018-08-09 ]

Yes.

Comment by Elena Stepanova [ 2018-08-12 ]

Is it anyhow different from MDEV-16923, so that you wanted to report it separately?

Comment by Sander Pilon [ 2018-08-12 ]

As we've rolled back all servers to 10.1, I cannot give anymore meaningful feedback.

Comment by Elena Stepanova [ 2018-08-12 ]

I will then assume that it's the same issue as MDEV-16923.

Generated at Thu Feb 08 08:32:34 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.