Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-12650

option 'server_id': unsigned value 0 adjusted to 1

Details

    Description

      When starting 10.2 with server-id=0 or without server-id configured, following note is printed:

      2017-05-01 8:26:22 140643309708800 [Warning] option 'server_id': unsigned value 0 adjusted to 1

      And the value indeed becomes '1'

      MariaDB [(none)]> show variables like 'server_id';
      +---------------+-------+
      | Variable_name | Value |
      +---------------+-------+
      | server_id     | 1     |
      +---------------+-------+
      

      According to documentation '0' is valid value for server_id indicating that this server is not intended to be part of replication.

      Thus either 0 should preserve as valid value or documentation must be changed.

      Attachments

        Issue Links

          Activity

            I don't mind having server_id=1 by default; we just need to update documentation and probably provide instructions about steps which make sure that particular instance is not part of replication (if DBA want that and we want such feature) https://github.com/MariaDB/server/commit/1c91569 , (plus fix MDEV-12597)

            anikitin Andrii Nikitin (Inactive) added a comment - I don't mind having server_id=1 by default; we just need to update documentation and probably provide instructions about steps which make sure that particular instance is not part of replication (if DBA want that and we want such feature) https://github.com/MariaDB/server/commit/1c91569 , (plus fix MDEV-12597 )

            Please find out where the change came from. I remember monty wanted to change the default, and it made sense, but I don't believe that's quite what was it meant to be. I think it should be the hard default that's changed – it makes no sense to set it to 0, then complain about it and adjust it to 1. And if it's set to 0 explicitly, it should be preserved.

            elenst Elena Stepanova added a comment - Please find out where the change came from. I remember monty wanted to change the default, and it made sense, but I don't believe that's quite what was it meant to be. I think it should be the hard default that's changed – it makes no sense to set it to 0, then complain about it and adjust it to 1. And if it's set to 0 explicitly, it should be preserved.

            In provided commit and current tree indeed I see that default is adjusted to 1 as well, but probably it gets overwritten somewhere else:

            -       VALID_RANGE(0, UINT_MAX32), DEFAULT(0), BLOCK_SIZE(1), NO_MUTEX_GUARD,
            +       VALID_RANGE(1, UINT_MAX32), DEFAULT(1), BLOCK_SIZE(1), NO_MUTEX_GUARD,
            

            anikitin Andrii Nikitin (Inactive) added a comment - In provided commit and current tree indeed I see that default is adjusted to 1 as well, but probably it gets overwritten somewhere else: - VALID_RANGE(0, UINT_MAX32), DEFAULT(0), BLOCK_SIZE(1), NO_MUTEX_GUARD, + VALID_RANGE(1, UINT_MAX32), DEFAULT(1), BLOCK_SIZE(1), NO_MUTEX_GUARD,

            It is not printed for me " without server-id configured". You probably have this server_id=0 in some my.cnf .Could you check?

            wlad Vladislav Vaintroub added a comment - It is not printed for me " without server-id configured". You probably have this server_id=0 in some my.cnf .Could you check?

            Yeah you are right - my fault - somehow I overlooked that dealing with many issues at the same time. I've asked Ian to address this in 10.2 documentation, will close the bug once documentation is fixed

            anikitin Andrii Nikitin (Inactive) added a comment - Yeah you are right - my fault - somehow I overlooked that dealing with many issues at the same time. I've asked Ian to address this in 10.2 documentation, will close the bug once documentation is fixed
            greenman Ian Gilfillan added a comment -

            The change of default has been documented

            greenman Ian Gilfillan added a comment - The change of default has been documented

            People

              greenman Ian Gilfillan
              anikitin Andrii Nikitin (Inactive)
              Votes:
              0 Vote for this issue
              Watchers:
              4 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.