Multiple Installation Issues - 0.5.0 - Vanilla (k3s) Kubernetes

I’ve re-setup my k3s Kubernetes cluster and had some time today to get Gitpod reinstalled. Unfortunately without any documentation it was quite a pain in the ass and it still doesn’t work properly.

The ws-daemon workload fails because there’s no ws-daemon Docker image at eu.gcr.io/gitpod-io/self-hosted/ws-daemon:0.5.0. Any ideas on why there is no ws-daemon Docker image (anywhere)?

The ws-proxy workload fails with this error in the log:

{"@type":"type.googleapis.com/google.devtools.clouderrorreporting.v1beta1.ReportedErrorEvent","error":"config validation error:
    github.com/gitpod-io/gitpod/ws-proxy/cmd.getConfig
        /tmp/build/components-ws-proxy--app.ca3b7dcc9d8c3c1634ff4e7f1297c63007ce8d00/cmd/root.go:203
  - TheiaServer not configured:
    github.com/gitpod-io/gitpod/ws-proxy/pkg/proxy.(*TheiaServer).Validate
        /tmp/build/components-ws-proxy--app.ca3b7dcc9d8c3c1634ff4e7f1297c63007ce8d00/pkg/proxy/config.go:102","filename":"/config/config.json","message":"cannot load config","serviceContext":{"service":"ws-proxy","version":""},"severity":"fatal","time":"2020-12-02T03:28:11Z"}

What can be done to resolve that one?

Also, any ETA on when new documentation will be up?

Sorry, @astuffedtiger that the state of deploying Gitpod on Vanilla Kubernetes is not that mature yet. It’s still in an early state in we currently working hard on getting a new version (ETA January 2021) with proper documentation on the road.

That said, it seems that you use helm charts from master with a Gitpod version 0.5.0. That does not work because the current helm charts use new components that do not exist for version 0.5.0. You need to take helm charts that match to Gitpod 0.5.0. Haven’t tested it but these should work: https://github.com/gitpod-io/gitpod/tree/357dc250e1a6666c53a94ca855903f3e52a217ee/chart

1 Like

Hi @corneliusludmann! Thanks for the follow up! So if I checkout that commit, and edit values.yml do I change the version in there to 0.5.0? Do I leave the imagePrefix as is or change it to eu.gcr.io/gitpod-io/self-hosted/? I assume when I’m all done editing that file I could run:

helm upgrade --install gitpod -f values.yaml -n gitpod .

Thanks!

hi

if you check my post history and on github too it should be enough to get you sorted

in summary

use

version: 0.5.0
imagePrefix: eu.gcr.io/gitpod-io/self-hosted/

set up your own mysql and minio instance as well as a seperate external registry

helm install gitpod gitpod/gitpod --timeout 60m --values values.yaml

then you may need to do a host of other things such as fixing extensions issue, fixing minio issues, ssl certs, patching your proxy, creating secrets etc etc

i was in the process of writing a full guide, but with @corneliusludmann estimating a new release in 2021 I think will hold back

it is do-able though but does take a lot of tweaking. I do believe collectively between us we have managed to get a working install of this, u just need to do a lot of reading here and on github gitpod site

Hi @hm2075, I actually did see your post on here and used that while getting what I did have setup, setup. I already have my own MySQL, MinIO and Docker Registry. It would be awesome if there was a guide to get this working right now, perhaps I will do it myself as reading from many different posts on here and GitHub is not fun. It would be better to have it all in one place.

Did you need to use that commit that @corneliusludmann mentioned or did you use master?

I don’t have to do any proxy patching as MetalLB works great.

Hopefully I can find some time later today to try this out again.

yes

you could clone then edit

let us know how u get on,

u could in theory use

imagePrefix: eu.gcr.io/gitpod-core-dev/build/

and for version use one of the later master branches, but it wont be supported by gitpod team

best thing, stick to self-hosted a v0.5.0

Still having issues, this time with imageBuilder:

helm upgrade --install gitpod gitpod/gitpod --timeout 60m --values values.yaml -n gitpod

Error: UPGRADE FAILED: failed to create resource: Secret "image-builder-registry-secret" is invalid: data[.dockerconfigjson]: Invalid value: "<secret contents redacted>": unexpected end of JSON input
  imageBuilder:
    name: "image-builder"
    dependsOn:
    - "image-builder-configmap.yaml"
    hostDindData: /var/gitpod/docker
    registry:
      name: registry.cypress.linuxbox.ninja/gitpod
      secretName: image-builder-registry-secret
      path: "secrets/registry-auth.json"

My secrets/registry-auth.json file is definitely valid JSON. I am running my helm install command from the gitpod/charts/ directory.

Welp, I’ve made it a bit further, but it looks like there’s an issue with Theia causing the workspace to fail to create. It sounds like I may not be the only one experiencing this issue. I see this in the workspace pod logs which may or may not be related:

{"message":"shutting down health endpoint","serviceContext":{"service":"theia-supervisor","version":""},"severity":"info","time":"2020-12-03T04:01:45Z"} 
{"message":"Theia supervisor loop ended - waiting for Theia to come down","serviceContext":{"service":"theia-supervisor","version":""},"severity":"info","time":"2020-12-03T04:01:45Z"} 

{"error":null,"message":"Theia was stopped","serviceContext":{"service":"theia-supervisor","version":""},"severity":"warning","time":"2020-12-03T04:01:49Z"} 

Any ideas?

Ok, I was able to fix this issue with Thiea Server. I’m having issues uploading extensions still, but I already have a thread on that. I will update that thread.