[MDEV-27846] 10.5.12, 10.5.13, 10.5.15 crashes on int wsrep::client_state::ordered_commit(): Assertion `owning_thread_id_ == wsrep::this_thread::get_id()' failed. Created: 2022-02-15 Updated: 2023-04-11 Resolved: 2023-04-11 |
|
| Status: | Closed |
| Project: | MariaDB Server |
| Component/s: | wsrep |
| Affects Version/s: | 10.5.12, 10.5.13, 10.5.15 |
| Fix Version/s: | N/A |
| Type: | Bug | Priority: | Major |
| Reporter: | Sergey Sokolov | Assignee: | Jan Lindström (Inactive) |
| Resolution: | Not a Bug | Votes: | 0 |
| Labels: | None | ||
| Environment: |
Debian 11 bullseye, mariadb and galera installed from mariadb repository. First time this crush was happened with 10.5.12 from debian repository, two upgrades to 10.5.13 and 10.5.15 from mariadb-repo also dont make us happy. |
||
| Attachments: |
crush_log.txt
full-mysql-error.log.gz
print_defaults
variables.txt
|
| Description |
|
Master-master cluster has two nodes - one main node and one in hot reserve. Switching between nodes doing with keepalived daemon (VRRP). Crashes are unpredictable: it can work about week and crush in the middle of work day. Also it can crush immediatly after IST. Unlike |
| Comments |
| Comment by Sergey Sokolov [ 2022-02-15 ] |
|
In full error log you can see all history of this crushes on various versions of mariadb-server and galera. Crush always happens with stack trace from crush_log.txt. |
| Comment by Sergey Sokolov [ 2022-02-28 ] |
|
Hello. We had low number of open files while our server crushes, innodb_open_files was set to 512. Now server 10.5.15 works without crushes more than 2 weeks with innodb_open_files=32768. I don't know - this is a bug or a feature, and documentation can't answer to this question. Thank you. |