addition of signadot job submit -f ... --wait

[scott-cotton]

24-12-30 scott@air cli % cat ~/tmp/job.yaml
spec:
  namePrefix: jwait
  runnerGroup: scott-jrg
  script: |
    #!/bin/sh
    echo not exiting-28
24-12-30 scott@air cli % ./signadot job submit -f ~/tmp/job.yaml --wait
Job jwait-lfl0zyi succeeded on Job Runner Group: scott-jrg

Dashboard page: https://app.signadot.com/testing/jobs/jwait-lfl0zyi

24-12-30 scott@air cli % echo $?
0

24-12-30 scott@air cli % cat ~/tmp/job.yaml
spec:
  namePrefix: jwait
  runnerGroup: scott-jrg
  script: |
    #!/bin/sh
    echo exiting-28
    exit 28
24-12-30 scott@air cli % ./signadot job submit -f ~/tmp/job.yaml --wait
Job jwait-k3mbgby failed on Job Runner Group: scott-jrg

Dashboard page: https://app.signadot.com/testing/jobs/jwait-k3mbgby

Error: job "jwait-k3mbgby" failed
24-12-30 scott@air cli % echo $?
1

24-12-30 scott@air cli % ./signadot job submit -f ~/tmp/job.yaml --wait --attach
Error: cannot specify both --attach and --wait

[davixcky]

Just to get a little bit of context, the main difference is that wait don't print the logs? Or there is other difference?

[scott-cotton]

Just to get a little bit of context, the main difference is that wait don't print the logs? Or there is other difference?

The main objective of this is to make it simple to run a job and determine in CI (automatically) whether it succeeded. --attach has the disadvantage that, due to retries, the output isn't parseable or really usable and makes it awkward at best to retrieve the job name. Here, one can also use -o json | jq -r .name to get the name of the job for subsequent artifact downloads.

[foxish]

Overall seems like a good idea and the fact that it's orthogonal to --attach makes sense.

[foxish]

In sb apply, we have --wait and --wait-timeout. Here, we have this timeout property that looks sort of related to wait. I'm curious whether this timeout is the equivalent of --wait-timeout when using with --wait?

[scott-cotton]

yes, it is. I guess we could add an alias or something to make it more uniform

[scott-cotton]

I double checked this and

1. I corrected the description, which was improperly worded before this PR, now it applies to both --attach and --wait
2. I fixed how --timeout applies to --wait; I had tested that but I guess I got the test wrong or it was on a dev version. Now it actually checks the timeout when non-zero during the --wait poll.

[daniel-de-vera]

LGTM, just left a minor comment.