Unable to deploy some GKE jobs because the name was too long
Bug #1825493 reported by
Po-Hsu Lin
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
ubuntu-kernel-tests |
Fix Released
|
High
|
Sean Feole |
Bug Description
Several qrt-kernel-
- Invalid value for field 'resource.name': 'b-lgke-
I think it's because the name was too long.
Changed in ubuntu-kernel-tests: | |
status: | Confirmed → Triaged |
To post a comment you must log in.
Yea, i'm seeing this carry over into azure, Since we dropped many non-critical elements of the job name, I see this much less of a problem. Only in particular cases.
For Example:
Azure has ridiculously long instance names which are appended to our instance job name, this in addition to testing with an azure-edge kernel will yield a somewhat similar result.
I don't want to redesign the job naming method in RegressionTestJ obsFactoryBase. jobname. I'm thinking to add special cases, where in the event we are testing with a -edge package, we simply drop the package or the kernel version from the cloud job name...
So...
b-lgke- 4-15-gke- 4-15-4- 15-0-n1- highcpu- 16-qrt- kernel- aslr-collisions
would become....
b-lgke- 4-15-gke- 4-15-n1- highcpu- 16-qrt- kernel- aslr-collisions
Again, this is a special case, because our package includes the '-4.15' as a part of the name. None of our other packages do that.
I'll own this and tackle it in the next slew of updates.