4 Apr
2023
4 Apr
'23
7:52 a.m.
On 4/3/23 23:48, Ian Wienand wrote:
I am guessing it was done like this to serialise the volume release and avoid having multiple "vos release" operations in flight at once. It seems like we could rework this now to use a semaphore, and have each platform run its own release job and have Zuul make sure they don't run together? This could be done as a job graph that reduces down to a vos release job.
Or you could use the new playbook semaphore feature. Put the vos release in its own playbook and scope the semaphore to just that. That probably makes sense in a post-run playbook, so this doc link: https://zuul-ci.org/docs/zuul/latest/config/job.html#attr-job.post-run.semap... Lots of choices now. :) -Jim