• Icon: Bug Bug
    • Resolution: Unresolved
    • Icon: Major Major
    • mercurial-plugin
    • None

      Mercurial polling on a Windows slave reports there are no changes even when there are.

      The error reported is different from JENKINS-11222, so, I created a new ticket.

      Jenkins reports http authorization missing, and no changes are seen. If I kick off the build manually, no issues with authorization, as well as the changes are brought in.

      Mercurial Polling Log:

      Started on Aug 8, 2012 8:00:05 PM
      [****] $ hg pull --rev default
      abort: http authorization required
      [****] ****
      Done. Took 0.79 sec
      No changes

      Is there a problem polling with http access?

      Jenkins ver. 1.461
      Hg: 1.38

          [JENKINS-14734] Jenkins Polling fails on windows slaves

          Ken Morse added a comment -

          I see the same behavior.

          Mercurial Polling Log

          Started on Sep 23, 2013 3:27:05 PM
          Polling SCM changes on buildvpc8
          [app] $ hg pull --rev default
          abort: http authorization required
          [app] $ hg log --rev default --template

          {node}

          [app] $ hg log --rev default --template

          {rev}

          Done. Took 0.39 sec
          No changes

          "Build Now" works just fine.

          Ken Morse added a comment - I see the same behavior. Mercurial Polling Log Started on Sep 23, 2013 3:27:05 PM Polling SCM changes on buildvpc8 [app] $ hg pull --rev default abort: http authorization required [app] $ hg log --rev default --template {node} [app] $ hg log --rev default --template {rev} Done. Took 0.39 sec No changes "Build Now" works just fine.

          Ken Morse added a comment -

          Is there any additional logging that could be turned on to try to sort out this problem?

          Ken Morse added a comment - Is there any additional logging that could be turned on to try to sort out this problem?

          Ken Morse added a comment - - edited

          Figured this one out – it turns out the first check (the one in red in the comment above) is done on the master, the rest is done on the slave. This explains how the first check can fail and the rest succeed in the same poll.

          We run both Jenkins master and slaves as Windows services. After an upgrade, our Jenkins master service was reset to logging in as the "Local System account" rather than the user account that had the password for Hg. Once we switched the Log on for the service back to the appropriate account, the initial polling succeeded again.

          Ken Morse added a comment - - edited Figured this one out – it turns out the first check (the one in red in the comment above) is done on the master , the rest is done on the slave . This explains how the first check can fail and the rest succeed in the same poll. We run both Jenkins master and slaves as Windows services. After an upgrade, our Jenkins master service was reset to logging in as the "Local System account" rather than the user account that had the password for Hg. Once we switched the Log on for the service back to the appropriate account, the initial polling succeeded again.

            kohsuke Kohsuke Kawaguchi
            vasubhat Vasudeva Bhat
            Votes:
            1 Vote for this issue
            Watchers:
            4 Start watching this issue

              Created:
              Updated: