Worker models error

[ygwane]

Hi,

I need help to solve an issue related to worker models.
We use CDS 0.48 release version on fresh install environment.
Workers are spawned on a swarm hatchery launched on remote server and CDS is launched with docker-compose, the registration of worker model seems to be good but worker model is in error.

Console output on worker model into CDS UI, display this (no specific error) :
p2021-03-01T18:34:18.394199372Z % Total % Received % Xferd Average Speed Time Time Time Current n2021-03-01T18:34:18.394241759Z Dload Upload Total Spent Left Speed 2021-03-01T18:34:18.472359704Z 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0 100 19.8M 100 19.8M 0 0 254M 0 --:--:-- --:--:-- --:--:-- 254M S2021-03-01T18:34:18.488106774Z 2021-03-01 18:34:18 [32m[INFO] [0m Registering with Token eyJhbGciOiJS on http://192.168.0.100:8081 [32mcaller [0m github.com/ovh/cds/engine/worker/internal.(*CurrentWorker).Register [32msource file [0m /tmp/790ad7b9a1a85167593986ed564384b7/run/engine/worker/internal/register.go [32msource line [0m 15 )2021-03-01T18:34:18.575704514Z 2021-03-01 18:34:18 [32m[INFO] [0m Unregistering worker [32mcaller [0m github.com/ovh/cds/engine/worker/internal.(*CurrentWorker).Unregister [32msource file [0m /tmp/790ad7b9a1a85167593986ed564384b7/run/engine/worker/internal/register.go [32msource line [0m 66

Debug log on hatchery and no revelant error in it.
The only warning error that we can see is on CDS API:
[WARN] service.JWTMiddleware> invalid given jwt token [/auth/consumer/worker/signin]: crypto/rsa: verification error

FYI, same environment works perfectly with CDS 0.47 and we try to setup into 2 differents environments.

Thanks in advance for help
Gilles

[yesnault]

Hi @ygwane, on the worker model page, what's the value for the flag need_registration?

[ygwane]

Hi, flag is set to false

[yesnault]

ok, so it's not really a problem to use the worker model; This error occurs when the worker try to unregister himself from the api, we will check that. It's not a blocker, this model is fully usable.

[yesnault]

@ygwane on worker model page, above the Checklist, do you see a message beginning with Exec routines killAwolWorker killAndRemove CheckWorkerModelRegister ? If yes, can you copy/paste message here?

If you resave the worker model, the worker model should be registered again after few minutes. Do you see the error message agin on unregister?

[ygwane]

Yes, here is the message above checklist:
Exec routines killAwolWorker killAndRemove CheckWorkerModelRegister: internal server error (from: an error occurred when registering the model with container name: /register-youpy-hopeful-but-inspiring-swanson) (caused by: internal server error)

And yes too, if I resave the worker model, same message appears on worker model UI page and into the api log output