defang icon indicating copy to clipboard operation
defang copied to clipboard

CLI not detaching after deployment

Open raphaeltm opened this issue 1 year ago • 1 comments

I've got a project with a normal service (a Django app) as well as managed Redis and managed Postgres.

When I run defang compose up it never detaches, despite the three services being healthy: the django app shows up as green in the portal withSERVICE_DEPLOYMENT_COMPLETED and Redis and Postgres both show up with a black circle and UPDATE_QUEUED.

raphaeltm avatar Sep 05 '24 20:09 raphaeltm

Happened also in https://github.com/DefangLabs/s.defang.io/actions/runs/10782434867/job/29902537954

lionello avatar Sep 09 '24 23:09 lionello

Is this Playground or BYOC @raphaeltm ? From the reference to Portal, seems like Playground, but then didn't we disable managed Postgres and Redis in Portal?

Prakash-Sundaresan avatar Oct 16 '24 23:10 Prakash-Sundaresan

Is this Playground or BYOC @raphaeltm ? From the reference to Portal, seems like Playground, but then didn't we disable managed Postgres and Redis in Portal?

@Prakash-Sundaresan this was before we disabled it.

raphaeltm avatar Oct 16 '24 23:10 raphaeltm

Would it still repro is the question...

Prakash-Sundaresan avatar Oct 16 '24 23:10 Prakash-Sundaresan

Because we can't detect state of managed services, they never makes it to the "COMPLETED" state. May be we should just assume managed services are up and just depend on the completion state of the other services.

nullfunc avatar Oct 18 '24 16:10 nullfunc

Isn't there an AWS event for RDS deployment complete?

Prakash-Sundaresan avatar Oct 18 '24 18:10 Prakash-Sundaresan

If not, there's got to be something we can do to poll? Given that we have the password, etc. we could run something like pg_isready (https://www.postgresql.org/docs/current/app-pg-isready.html) until it returns true or a timeout we set. With redis, we can run redis ping until we get pong back.

raphaeltm avatar Oct 18 '24 21:10 raphaeltm