How can we improve Azure DevOps?

Rerun failed build task/step

This suggestion is migrated to Developer Community. Please use below link to view the current status.
Provide a way/configuration people could be able to rerun some failed tasks server times in the build pipeline.

70 votes
Sign in
Password icon
Signed in as (Sign out)

We’ll send you updates on this idea

Patrick shared this idea  ·   ·  Flag idea as inappropriate…  ·  Admin →


Sign in
Password icon
Signed in as (Sign out)
  • Anonymous commented  ·   ·  Flag as inappropriate

    I was looking for the option to re-run failed tasks all over the settings of our build and then googled and ended here. Can't believe that it's something that's not supported.

  • Deepika commented  ·   ·  Flag as inappropriate

    sometimes a task fails on the first run, though there is no definite reason for it. But when ran again immediately it succeeds. So Retry as one of the control option would definitely help.

  • Scott Heath commented  ·   ·  Flag as inappropriate

    I know everything should work right the first time <grin> but I have an Azure ARM deployment that sometimes fails when running a custom extension. Doesn't seem to be a rhyme or reason to it as it will sometime go a month without failing and of the three machines it deploys normally only one will fail. Having the ability to re-run a deployment phase would be a huge help!

  • Anonymous commented  ·   ·  Flag as inappropriate

    This is one of the top asks on my team. Sometime some build steps fail due to silly reasons and its a huge waste of resource and dev time to re-queue a full build at that point. It would be great, if possible, to fix the issue and restart the build from the failed step, instead of kicking off a full new build.

  • Anonymous commented  ·   ·  Flag as inappropriate

    To circumvent having our build fail due to external dependencies, we've had to move tasks to task-groups which now contain the same task multiple times. This simulates the requested behaviour, but is really messy.

  • Huw commented  ·   ·  Flag as inappropriate

    This really should be added - so tedious trying to setup new build definitions if they're complicated and need a bit of trial and error to get working

  • Chris Wood commented  ·   ·  Flag as inappropriate

    Really need this; our deployments sometimes fail because of transient cloud issues beyond our control, and we need to be able to re-run a failed step n times or re-run a deployment n times.

  • DevMgr commented  ·   ·  Flag as inappropriate

    totally agree, this should be a standard feature really. Sometimes build fails for "silly" reasons and it's very time consuming to restart another build from step zero!!!

  • Angel commented  ·   ·  Flag as inappropriate

    Please make this happen soon. This is one of the key feature missing when you migrate from MTM to TFS. People cannot easily rerun failed test to do triage or investigation, or just need the unstable test to pass to make the test run green.

  • John Buxton commented  ·   ·  Flag as inappropriate

    This feature needs to be added quickly if Microsoft want to catch up with the competition.

    (Note that I did not say *keep* up with the competition, because Microsoft are already years behind on this and other things.)

  • Dave commented  ·   ·  Flag as inappropriate

    Yes please, It's impossible to incrementally design definitions at the moment! Very time consuming!

  • Victor commented  ·   ·  Flag as inappropriate

    This feature very necessary. After migration to TFS 2018 we cannot use the same feature from MTM.

  • Martti Kontula commented  ·   ·  Flag as inappropriate

    This is needed. Sometimes a build may fail due to external dependencies. It would be useful to retry the build with exactly same configuration by a click of a button. "Re-run".

  • Nick commented  ·   ·  Flag as inappropriate

    I'd tag this for release. We have a smoke test task that fails intermittently and it ends up failing the whole release. In order to turn it "green", the team has to completely re-run the release. But this takes significant time.

    I'd like to see an "Retry count" (default to zero) as one of the "Control Options" for any given task. If that task fails, it'll retry up to that many times and only fail the step if it never gets a success.

Feedback and Knowledge Base