Provide ability for user to indicate a stage should be skipped

Comments

3 comments

  • Avatar
    Yogi Kulkarni

    Hi Ian,


    This is an interesting use case, could you share more details about it and why a particular stage would need to be skipped.


    In the upcoming release we plan to add a Rerun-Job feature which could potentially handle some scenarios where you need to skip stages, e.g. given a deployment stage that deploys to multiple agents, if one of the deployments to an agent fails, currently you need to rerun the entire stage. Instead with Rerun-Job, you can run the one that failed.


    I just wanted to check if your user case is different.


    Thanks,


    Yogi

    0
    Comment actions Permalink
  • Avatar
    Ian Bridson

    Hi Yogi


    We have split our 3 release processes into 3 stages. The first is a .Net release to multiple agents on web and application servers (one job for each), then if this succeeds there is a refresh database job that sets db to current live code, and finally a build db that ftp db files and runs scripts to create latest db with PL/SQL packages. Now sometimes the database doesn't want to be refreshed and the PL/SQL applied on top of the last Pl/SQL, not from live. Therefore the second stage needs to be skipped.


    As mentioned above this is achieved by logic in the refreshdb script base don environment variale settings, but go still has to assign an agent and run this script for the stage to complete and move on to the build, tying up resources unnecesarily and wasting deployment time.


    Go should be able to skip this stage without any agent involvement.


    Your suggestion of rerun job sounds intereting for our First stage , deploy to multiple agents, as we do have occasions where one of 8 web servers fail and i have to roll back and rerun the stage, even though 7 agents deployed successfully. This option would save rerunning the app server job, but in production i am not sure how robust this would be in an operator deciding which jobs to rerun.


    Could the rerun job actually be rerun failed agent for job?

    0
    Comment actions Permalink
  • Avatar
    Anush Ramani

    Hi Ian,


    Yes, the proposed feature would in fact re-run the job on the same agent on which it had run on.

    0
    Comment actions Permalink

Please sign in to leave a comment.