Self hosted server with custom ingress failing to work as expected

[ala_eddine]

Hello speckle support,

I deployed the latest version of speckle locally on kubernetes using the Helm chart you provide, but i still get the old interface of speckle. my question is how to get the new speckle and deploy it locally ?

regards

[iainsproat]

It requires the following to be added to the values.yaml file:

frontend_2:
  enabled: true

Iain

[ala_eddine]

Hi @iainsproat

Thanks for your replay, i didn’t get the access to speckle on browser i got this error

and when back to the old frontend it works

[iainsproat]

Hi @ala_eddine

You will need to search the nginx, speckle-frontend-2, or even the speckle-server pods for additional logs which may help pinpoint the cause of this issue.

There’s unfortunately not enough information at the moment to identify the root cause of this issue and suggest a fix.

Iain

[ala_eddine]

hi @iainsproat

I’m using my own ingress file after made some modification the interface can be showen but unable to register an account ,

can you advise if it’s specific configuration on the ingress file
here’s my current configuration

[iainsproat]

If the registration failed, you may wish to inspect the browser console logs or view the nginx, frontend-2, or speckle-server pod logs for more information which may help with debugging the issue.

Iain

[ala_eddine]

Hi @iainsproat

i’m not using nginx as ingress i’m using traefik

the logs from browser console:

the logs from speckle-server:

the logs from speckle-frontend:

also my ingess configuration:

[iainsproat]

I think the frontend screenshot is a duplicate of the server screenshot, did you upload it twice by mistake?

It appears that the requests from the browser receive a 404 response (the screenshot doesn’t show the full status code), but we don’t see the corresponding requests in the server logs.

The server logs instead show 200 response http status codes (i.e. success) for requests to the /graphql endpoint. However, these are originating from within Kubernetes (due to the host being 127.0.0.1). These are likely part of Kubernetes’ pod readiness checks, so are unlikely to be the requests sent by your browser.

What do the traefik (or nginx) controller logs show?

Iain

[ala_eddine]

hi @iainsproat
I change the ingress to use the nginx which was enabled on the value file and here’s the logs from nginx:

also the logs from the frontend-2 pod on debug mode:

the logs from frontend-2 info mode:

[ala_eddine]

Hi @iainsproat

is there any suggestion about that , Thanks in advance.

kind regards

[iainsproat]

After changing to nginx was the ‘Registration failed’ error still occurring? And were the browser console errors still the same?

Iain

[ala_eddine]

Hi @iainsproat
Thanks for your response

yes the same error still occuring with nginx ingress

Note i’m using the latest speckle-server image

[iainsproat]

The nginx log shows the 404 being returned for the /auth/local/register endpoint. The nginx configuration should have routed it to the frontend-2 pod.
However, I don’t see that endpoint being requested in the frontend-2 pod logs. Can you search the logs for frontend-2 to see if /auth/local/register is being requested?

Iain

[ala_eddine]

Hi @iainsproat

Here’s from frontend-2 when click register :

and from speckle server:

[ala_eddine]

Also when enabling the nginx ingress from the values file an deploy i got this message:

[iainsproat]

Thanks for highlighting that a warning is generated by the ingress.

From reading nginx documentation it seems that the warning is invalid and can be ignored: "use-regex" Annotation Broken: "Warning: path /something(/|$)(.*) cannot be used with pathType Prefix" · Issue #10200 · kubernetes/ingress-nginx · GitHub

From the frontend-2 pod logs, it does appear that a request to /auth/local/register has been successfully handled by it. However, I cannot see a call to /auth/local/register?callback=... in the provided logs.

Iain

[ala_eddine]

Hi @iainsproat

i search about “/auth/local/register?callback” and i didn’t found it , there’s just what in the provided logs

[iainsproat]

Hi @ala_eddine

At the moment it feels like the errors are still a hangover from your earlier customisations.

Have you tried deploying the server on kubernetes from scratch following the documentation exactly without any customisation?

If that works for you, then I would use that as the base from which to start incrementally making small customisations if they are still required.

Iain

[ala_eddine]

Hi @iainsproat
on our deployment the only customisation was on the ingress, and i removed it and used the default nginx ingress, on our environement i use postgres, redis and minio all deployed on kubernetes
can you tell me what we have to check on the environement redis, postgres and minio and so on thanks in advance

Kind regards

[jonathon]

@ala_eddine

can you tell me what we have to check on the environement redis, postgres and minio and so on

Most everything you ask is specified in the Helm chart. Thats why it is there.

The list is quite extensive: https://specklesystems.github.io/helm/