Update the promotion test jobs to such that not all the tests need to pass in a parallel run
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
tripleo-quickstart |
Invalid
|
Undecided
|
Unassigned |
Bug Description
Not sure if this is the best place to track this issue, however, I'll continue.
The rdo master promotion job for example [1]. Durring the test phase (INSTALL / TEST (IMPORT IMAGES)) all the jobs in this section need to pass on the same run for a promotion to take place. This often makes it difficult to promote due to random infra issues that may fail a single job.
A better less error prone method would be to store the job status and hash used on an external file server and have each job write it's status in a file remotely.
Something like...
JOB: STATUS: DELOREAN HASH:
Job1 passed 23erfgy78uio0
Example of a promotion file that has yet to occur:
JOB: STATUS: DELOREAN HASH:
Job1 passed 23erfgy78uio0
Job2 failed 23erfgy78uio0
Job3 passed 23erfgy78uio0
Example of a promotion file that is ready to be promoted:
JOB: STATUS: DELOREAN HASH:
Job1 passed 23erfgy78uio0
Job2 failed 23erfgy78uio0
Job3 passed 23erfgy78uio0
The current promotion job "rdo-promote-
This would allow us to run a failing job.. e.g. Job2 by itself w/o also having to kick off all the jobs.
[1] https:/
Changed in tripleo-quickstart: | |
status: | New → Invalid |