RunJobs check contstraint failed

[umglurf]

Describe the bug

Since boundary 0.2.3 I've started getting these errors in the controller logs:

[ERROR] controller: error getting jobs to run from repo: error="job.(Repository).RunJobs: db.DoTx: job.(Repository).RunJobs: wt_private_id_check constraint failed: check constraint violated: integrity violation: error #1000"

I upgraded to 0.3.0 hoping the issue was solved, but still the same. I then deleted the database and recreated it, but still the error occurs.

To Reproduce
Steps to reproduce the behavior:

1. Start boundary-controller
2. See error

Expected behavior

Additional context
Running with 3 boundary controllers and 4 workers. 3 workers on the same servers as the controllers, and one worker on another server.

[louisruch]

Hi @umglurf
Thank you for trying Boundary! I noticed this issue when running boundary dev and discovered it was a race between the server upsert and the new job scheduler requesting work. I resolved this in a wip branch with other work, but I forgot to merge this fix into main - sorry about that. However, since you are not running Boundary in dev mode it might not be the same race I found, do you see this error consistently or only once when starting the server?

[umglurf]

Hi

I see it consistently, not only when starting the server. It's been logged about 50 times per server since the last restart one hour ago.

[louisruch]

@umglurf 50 times in an hour is about what I would expect from this, you mentioned that you have 3 controllers are you providing the name in the controller stanza. If so can you please confirm what the three names are?

[umglurf]

controller 1:

name = "kara"

controller 2:

name = "sharon"

controller 3:

name = "laura"

[louisruch]

Thank you!

There is a bug in the Postgres job_run table where we are incorrectly enforcing a check constraint on the length on the controller name to be greater than 10. I have a PR that should resolve this issue.

[umglurf]

Great, hoping this will make it into 0.3.1