We have a few plans, all of which contain multiple stages which in turn contain a number of jobs and tasks. For example:
Plan 1
* Stage 1
* Stage 2
Plan 2
*Stage 1
* Stage 2
When Plan 1: Stage 1 is running, if another Plan is triggered, Plan 2: Stage 1 is queued, running before Plan 1: Stage 2.
So the run order becomes
Plan 1: Stage 1
Plan 2: Stage 1
Plan 1: Stage 2
Plan 2: Stage 2
We need currently running plan to finish before starting a new plan as (even though I know it's not recommended but our middleware requires it) there are artefacts shared between stages that may also be used/updated/modified by stages in another plan hence we cannot use artefacts between stages.
Without this, previous plans future stages can be corrupted before the next stages run, breaking out build process.
So ideally it needs to be
Plan 1: Stage 1
* Plan 2 triggered *
Plan 1: Stage 2
Plan 2: Stage 1
Plan 2: Stage 2
Is this possible without putting our entire plans steps into a single job with a lot of individual tasks because this would get rid of the stages that do have elements that can run in parallel?
Thanks Lee
Ps. Appologies for the formatting
Doesn't look like this is possible - quite a simple addition though I'd imagine
You can configure the plan to run as a dependency. Or rather configure your dependent build to be blocked if the other build is running. Take a look on the dependencies tab. There are a bunch of options there.
Block build if parent builds are queued or in progress If parent builds are queued or building, the current build will be blocked.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Thanks Gretchen that would be fine if I only had two plans on my build server (granted my description only had two plans as an example), but my build server had upwards of 20 plans, all of which I'd like to complete fully before kicking off any queued plans.
I've gotten around this by doing what I originally suggested myself (putting the entire plan contents in a single job) but it's not very managable or scalable.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.