Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Seandon Mooy
    @erulabs
    I see - deploy-node-app seems to have not created an Ingress for you
    And our UI isn't clear at all how to change an internal-only service into an internet-accessible one
    (the answer is to remove it and make a new, public one)
    Jeffrey Emerich Christian
    @15chrjef
    I get "Error Unexposing Port" when clicking x
    Seandon Mooy
    @erulabs
    Hrm - can you try refreshing your browser?
    I see that the service was removed properly
    but I think I might also see why the frontend gave you that message
    Jeffrey Emerich Christian
    @15chrjef
    @erulabs 1) ok, i'll check out making one public 2) is there a complete link anywhere like subdomain.username.cluster.kubesail.io:4000 where i can find the full link to access my frontend, backend, or anything that is publically exposed
    Seandon Mooy
    @erulabs
    Yes - it will show that address when you expose the port publicly
    We currently only support exposing HTTP/HTTPS traffic, but TCP is coming soon!
    (so graphql should work fine, but it will be https://graphql.15chrjef... etc without a port number)
    Seandon Mooy
    @erulabs
    I dont see any Ingress created for ya - I do see you deleted the service - you should now be able to expose that port to the internet
    Jeffrey Emerich Christian
    @15chrjef
    Got it, yes now i can expose the port as I would like!
    Zir
    @Zireael
    Hi, firstly, this chat is requesting excessive permissions when trying to log in with GitLab. Like, access to all my repos? Second, not sure this is the right place, but I don't see better way. I was testing Kubesail earlier today and when deleted a volume I was using, it is still showing up as 'Terminating' for the last couple of hours. The volume ID is pvc-7471859e-7748-43da-ba41-f8775ec5527c.
    image.png
    Zir
    @Zireael
    One of the issues I still need to figure out, is how to access my containers on GitLab when they're set as private. I guess I'll need to use an access token and set it up as a secret? Time to sleep, but if you know of a good guide to do that, I'd be grateful.
    Seandon Mooy
    @erulabs
    Hrm - sometimes volumes can be stuck Terminating if they are attached to a running Pod - I'll take a look at that volume shortly :)
    We're working on a tool in the UI to hint at creating ImagePullSecrets - check back soon :D
    Zir
    @Zireael
    Hi, I sat down this weekend to do some more KubeSail testing. Below are few of my comments - I hope this helps you in some way as feedback:

    1) If I expose port with a subdomain, ie https://subd.username.usw1.kubesail.io/ I get error NET::ERR_CERT_COMMON_NAME_INVALID
    Looks like your SSL certs aren't wildcard for subdomain. I have deleted TLS secret when playing around and now it doesn't get re-created.

    2) Deployments-> ports
    Try delete a port in UI. I get "Error unexposing port" every time. But after I refresh page, the port is gone and I can re-add it without problem

    3) It is not possible to see configmaps (env variables) in GUI, but it's possible to see secrets (https://kubesail.com/secrets). I have to go to kubectl and delete configmaps in command line. To delete secrets I can just use GUI on your website. Possible to show configmaps in your GUI?

    4) I have created two container solution (postgres + GraphQL). Got it woking now, however one issue I had was that POSTGRES_DB env variable was not being pulled from configmap. When I hardcoded the value, it works. Pulling env variables from Secrets works for me too. Not sure what happened there with configmap value, I'm just happy I got it to deply atm.

    You can see my entire project here:
    https://gitlab.com/zirdev/zir-docker

    You can run ./kubernetes/kubectl-deploy.bat to deploy to kubesail.com to check it out. If you expose external port for 'graphql', you should see a page at https://{username}.usw1.kubesail.io/graphiql but as per point 1), I'm getting SSL cert error if I expose on subdomain https://something.{username}.usw1.kubesail.io/graphiql

    snippet showing working/notworking env variables definitions:

            - name: POSTGRES_DB # works
              value: my_database
            # - name: POSTGRES_DB # doesn't work
            #   valueFrom:
            #     configMapKeyRef:
            #       key: POSTGRES_DB
            #       name: db-env
            - name: POSTGRES_USER # works
              valueFrom:
                secretKeyRef:
                  name: secret-kubernetes-postgresql
                  key: POSTGRES_USER

    5) My declarations for resources:requests:/limits: for CPU/RAM in yaml files (see https://gitlab.com/zirdev/zir-docker folder ./kubernetes/db-deployment.yaml and ./kubernetes/graphql-deployment.yaml) get ignored when I do 'kubectl create -f xxx-deployment.yaml'. Your GUI shows each of my deployemnts as 1core CPU and 16Mi RAM. If I edit yaml on your website through your GUI and add resources:requests:/limits: I get correct values displayed on your website.

    Do you have an email address? Or a way to DM - I wanted to send you a short private message.
    Zir
    @Zireael
    6) https://kubesail.com/templates <-- it is not possible to delete a template set if I create one by accident. Yes, I created one by accident if you're wondering :X
    7) Not sure if possible, but I was wondering whether you could allow doing deploys with docker-compose.yaml . Converting docker-compose with kompose bloats the whole thing out for me. But I guess I got used to it now. Next is trying out Helm I suppose...
    Seandon Mooy
    @erulabs
    Wow! Thank you so much for all your feedback - I'll go thru this with a fine tooth comb when I get home tonight
    We have huuuge plans and a mostly new UI coming soon - lots and lots of usability fixies
    I'll look into those bugs asap tho
    Thank you again, invaluable feedback <3
    (certs should work, there is a temporary cert issued which is not valid, but it should go valid quickly - typically it doesnt if the letsencrypt pod cant start (if you're out of resources))
    Ill double check your setup tho if you dont mind to see what happened there
    Zir
    @Zireael
    feel free to have a look - that's why I posted link to my repo (nothing personal/sensitive there, it's just a dry run for now). Feel free to fork and try out. I would say if you tinker with the repo, you might be able to add a stack of database + GraphQL server to your official templates.
    Zir
    @Zireael
    oops, I forgot to push commit to Github/Gitlab last night. Repo should be up to date with working stack now
    Zir
    @Zireael
    Lastly, a while ago I was reporting that a volume I was deleting was stuck as terminating. Have the same thing with a pod:
    image.png
    I've killed this pod around the same time as that volume. I can't tell you exactly what I did to make both hang - I was learning docker/kubernetes so probably had lots of bad code and tried random commands in kubectl :)
    Anyho, this pod isn't showing up on website GUI so I just discovered it chillin' there in my terminal.
    Zir
    @Zireael
    Not sure how exactly your service works, but maybe you could set up a cron job to go through pods/volumes and nuke any stuck Terminating for more than 24h?
    Seandon Mooy
    @erulabs
    Hrm, I'll take a look at the Terminating pods - they should eventually go away, but being stuck typically indicates some fault on a host somewhere
    Jacob Tomlinson
    @jacobtomlinson
    Can anyone help here. I'm trying to use the helm 3 beta with KubeSail (to avoid previous free tier resources being used up by tiller) but I'm getting the following error when I try to install my chart. Error: could not get apiVersions from Kubernetes: unable to retrieve the complete list of server APIs: metrics.k8s.io/v1beta1: the server is currently unable to handle the request
    Seandon Mooy
    @erulabs
    Hrm - that's not good - that implies our metrics service is unavailable - this sometimes happens when it's "flapping" - I'll take a look... my monitoring should have caught this!
    Seandon Mooy
    @erulabs
    Okay, this should be all resolved on the cluster you're on currently. Going to add some more specific monitoring to the metrics service - It's not technically part of what we consider mission critical, but it does seem like more and more tools rely implicitly on it.
    Jacob Tomlinson
    @jacobtomlinson
    Still getting the issue. (I'm using an account called opsdroid-bot not this one.)
    Jacob Tomlinson
    @jacobtomlinson
    Actually it looks like a slightly different error this time Error: could not get apiVersions from Kubernetes: unable to retrieve the complete list of server APIs: admission.certmanager.k8s.io/v1beta1: the server is currently unable to handle the request
    Seandon Mooy
    @erulabs
    Hrm... I don't see any resources in your namespace - I'll search for that error but possibly you can share the steps to get to that error? Applying a particular bit of YAML?
    will make sure the admission controller is healthy, but it appears to be healthy at first glance...
    Seandon Mooy
    @erulabs
    that error might be legitimate if the certificate you're requesting has an invalid hostname for example... Ill try to dig up the logs
    Jacob Tomlinson
    @jacobtomlinson
    I'm trying to install this help chart with the helm 3 beta https://github.com/opsdroid/helm-chart
    James Dibble
    @dibble-james
    Anyone had any issues verifying a custom domain where the DNS is hosted by Cloudflare? I've verified the txt records exist using mxtoolbox but I seem to get stuck in verifying attempts.
    Seandon Mooy
    @erulabs
    Hey @dibble-james - can you direct message me the name of the domain you're having trouble with? I'd be happy to take a look.
    I haven't heard any reports about that specifically, and I have a test domain via cloudflare that works - but I'd bet it's probably our bug and not something you've done wrong heh.