Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-16656

DROP DATABASE crashes Galera Cluster (all nodes)

Details

    Description

      Hi,

      I tried to delete a database from the cluster when the entire cluster suddenly crashes due to a DROP DATABASE operation. The crash appeared on all nodes at the same time as u can see in the logfiles.

      I added all logfiles and our my.cnf which is the same on all three nodes.

      If you have any advices for our configuration, please let me know.

      Best Regards,

      Aljoscha

      Attachments

        1. big-galera-01_crash.log
          6 kB
        2. big-galera-02_crash.log
          8 kB
        3. big-galera-03_crash.log
          8 kB
        4. my.conf
          2 kB

        Activity

          danblack Daniel Black added a comment - - edited

          Do you have a list of table names in that database? In particular is there a table name that forms a extension of the database name, like bob and bobs_friends?

          danblack Daniel Black added a comment - - edited Do you have a list of table names in that database? In particular is there a table name that forms a extension of the database name, like bob and bobs_friends ?

          There are and were no tables in the database, because the project never started and so I wanted to clean up the leftovers.

          MariaDB [(none)]> use example-net-live;
          Database changed
          MariaDB [example-net-live]> show tables;
          Empty set (0.00 sec)
          

          Horrorhorst Aljoscha Dembowsky added a comment - There are and were no tables in the database, because the project never started and so I wanted to clean up the leftovers. MariaDB [(none)]> use example-net-live; Database changed MariaDB [example-net-live]> show tables; Empty set (0.00 sec)
          Horrorhorst Aljoscha Dembowsky added a comment - - edited

          Another day another db...

          Today the whole cluster just locked the query execution while a script tried to drop another database. This time the database was filled with data but still no tables formed like the db name. I had to stop the loadbalancer and kill all threads followed by quick service restart on the nodes. Afterwards the applications went online again.

          DB-Name: example-com-dev
          Tables:

          wp_*
          user_*
          

          (* as wildcard)

          I will review the logs later and append them to the files above

          Horrorhorst Aljoscha Dembowsky added a comment - - edited Another day another db... Today the whole cluster just locked the query execution while a script tried to drop another database. This time the database was filled with data but still no tables formed like the db name. I had to stop the loadbalancer and kill all threads followed by quick service restart on the nodes. Afterwards the applications went online again. DB-Name: example-com-dev Tables: wp_* user_* (* as wildcard) I will review the logs later and append them to the files above

          sysprg Can you take a look at this during the next 10.1 sprint?

          ratzpo Rasmus Johansson (Inactive) added a comment - sysprg Can you take a look at this during the next 10.1 sprint?

          People

            sysprg Julius Goryavsky
            Horrorhorst Aljoscha Dembowsky
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved:

              Git Integration

                Error rendering 'com.xiplink.jira.git.jira_git_plugin:git-issue-webpanel'. Please contact your Jira administrators.