-
Improvement
-
Resolution: Unresolved
-
Major
-
None
The current implementation triggers polling on all jobs with Perforce Triggered Build enabled. This is fine for small installations with few check ins per day, but not for installations with hundreds of jobs. This can cause massive spikes in size increases for Perforce server logs.
For example, if I check in a file to "//my_depot/..." I should trigger polling on a job with view:
//my_depot/... //workspace_name/...
However, I should not trigger polling on a job with view:
//my_depot2/... //workspace_name/...
Additionally, it should verify the change is actually associated with the right port. That overhaul is possibly out of scope of this improvement request.
You might be able to accomplish that portion naively in the same request. If you retrieve the change in probeJobs using port and change you likely have a very high success ratio of matching the change to the server. If change is null, it's obviously a mismatch. If change isn't, then your depot path still has to match the jobs workspace view. I would be surprised if many Perforce deployments have large amounts of duplicated full depot paths on multiple servers.