Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-26706

Enable per table gtid domain_id

Details

    • Task
    • Status: Open (View Workflow)
    • Major
    • Resolution: Unresolved
    • None
    • Replication
    • None

    Description

      Replication filters have many issues and domain filtering works better to allow selective filtering on replicas such task would attach one or many domain per table definition enabling user define logical dataset to be filtered or apply on a replica. Today using extra domain is.an application layer task and is out of the scope of sysops dbops making such good feature a no go in most case.

      Attachments

        Issue Links

          relates to

          Task - A task that needs to be done. MDEV-6901 [PATCH] session.gtid_domain_id should be able to be set by non-SUPER user

          • Major - Major loss of function.
          • Stalled

          Task - A task that needs to be done. MDEV-4831 Replication to store filter in master.info

          • Minor - Minor loss of function, or other problem where easy workaround is present.
          • Open

          Bug - A problem which impairs or prevents the functions of the product. MDEV-26632 multi source replication filters breaking GTID semantic

          • Major - Major loss of function.
          • Closed

          Activity

            Ascending order - Click to sort in descending order
            danblack Daniel Black added a comment -

            This will get tricky if two table updates are in the same transaction. Which gtid will it get?

            danblack Daniel Black added a comment - This will get tricky if two table updates are in the same transaction. Which gtid will it get?
            stephane@skysql.com VAROQUI Stephane added a comment - - edited

            Yep in row based no issue but in statement could just works as in filters today so if stmt cross multi domains tables , all domains has to be store in binlog so that filetring can works as described today
            If the filter take place on master anyway on the request of a replica for saving bandwidth this does not need extra binlog domain information , if taking place on the slave the per table domain information can be found in table definition

            stephane@skysql.com VAROQUI Stephane added a comment - - edited Yep in row based no issue but in statement could just works as in filters today so if stmt cross multi domains tables , all domains has to be store in binlog so that filetring can works as described today If the filter take place on master anyway on the request of a replica for saving bandwidth this does not need extra binlog domain information , if taking place on the slave the per table domain information can be found in table definition
            stephane@skysql.com VAROQUI Stephane added a comment -

            For resolving the selective purge on master and store on replica a syntax ALTER TABLE ADD DOMAIN x FOR ALL|DELETE|UPDATE|REPLACE|INSERT can be introduce

            stephane@skysql.com VAROQUI Stephane added a comment - For resolving the selective purge on master and store on replica a syntax ALTER TABLE ADD DOMAIN x FOR ALL|DELETE|UPDATE|REPLACE|INSERT can be introduce

            People

              Unassigned Unassigned
              stephane@skysql.com VAROQUI Stephane
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

              Dates

                Created:
                Updated:

                Git Integration

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