A workflow's execution can take multiple branches simultaneously, and some nodes could have already traversed their "failure" or "always" connections. The best feature spec I could imagine would be to start again on all filed nodes that ended their branch. There's appeal as an analog to retry on failed hosts for job templates, but I think it would be difficult to communicate its function.

Any update on this feature?

Another way to do it is to leave it up to users. Select the template to start at and run from there with the understanding that any variables that were set in previous jobs would not exist anymore and that some branches could have already been passed. This way operators of the system can decide which is the better option for the specific workflow, start from the beginning or continue.

+1 On this feature. Definitely must have in roadmap.

After solving all necessary problems engineer goes to specific failure node job(linked with wf) and relaunches it, thus continuing workflow execution.

When is this getting released?

@spsingh1982 whenever somebody writes the code 😉

I think giving a 👍 on the Idea will be better than a comment with +1. We can avoid mail storm :-)

Is it on the roadmap? Really need this feature as well.

I'm seeing interest in this from large organizations with multi-hour workflows. +1