← Back to team overview

maria-developers team mailing list archive

Re: [Commits] 608b0ee52ef: MDEV-23033: All slaves crash once in ~24 hours and loop restart with signal 11

 

Hello Jan,

Please find the detailed explanation here

https://jira.mariadb.org/browse/MDEV-21153?focusedCommentId=176249&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-176249

Thank you

S.Sujatha

On 04/01/21 3:47 pm, Jan Lindström wrote:
Hi Sujatha,

On Mon, Jan 4, 2021 at 11:45 AM sujatha <sujatha.sivakumar@xxxxxxxxxxx <mailto:sujatha.sivakumar@xxxxxxxxxxx>> wrote:


    +    for (TABLE_LIST *tables= rgi->tables_to_lock; tables;
    +         tables= tables->next_global)
    +    {
    +      if (slave_run_triggers_for_rbr)
           {
             tables->trg_event_map= new_trg_event_map;
             lex->query_tables_last= &tables->next_global;
           }
    +      else if (!WSREP_ON)
    +      {
    +        tables->slave_fk_event_map= new_trg_event_map;
    +        lex->query_tables_last= &tables->next_global;
    +      }


Why is prelocking not done when WSREP_ON ? (or WSREP(thd) ) In my opinion it should be reasoned here with comment, many of our customers actually use Galera as master-slave setup. Did you test your testcase with Galera setup and did you check Galera test results on bb ?
Regards,
  Jan

References