by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Peter Salomonsen
    @petersalomonsen
    I just upgraded my account but still my pod won't start because of quota exceeded. (namespace petersalomonsen). Does it take some time to take effect?
    Borozenets
    @Borozenets
    Hey, I have the same problem.
    progressbarteam
    @progressbarteam
    Same here
    MO
    @8Mobius8

    Interestingly I may be seeing a similar problem in my Free space.

    So I tried emptying my space of and everything that I could. Verified that no pods are running that I can see in my space.

    I tried running kubernetes describe quota to see my usage and it still saying that i am using cpu on limit and requests. Not sure if that is suppose to happen or not.

    K8s experts what you think?

    $ kubectl describe quota
    Name:                         8mobius8-quota
    Namespace:                    8mobius8
    Resource                      Used   Hard
    --------                      ----   ----
    count/certificates            0      10
    count/configmaps              0      50
    count/cronjobs.batch          0      10
    count/daemonsets              0      0
    count/deployments.apps        3      10
    count/deployments.extensions  0      10
    count/ingresses               0      10
    count/jobs.batch              0      10
    count/persistentvolumeclaims  0      10
    count/pods                    1      10
    count/replicasets             0      20
    count/replicationcontrollers  0      10
    count/secrets                 5      25
    count/services                0      20
    count/services.loadbalancers  0      0
    count/services.nodeports      0      10
    count/statefulsets.apps       0      5
    limits.cpu                    1      1
    limits.ephemeral-storage      0      1Gi
    limits.memory                 100Mi  512Mi
    requests.cpu                  20m    20m
    requests.memory               100Mi  512Mi
    requests.storage              0      100Mi
    Seandon Mooy
    @erulabs
    Hello everyone! Sorry for the radio silence - this is a very tricky one - but it's our issue (unfortunately!) - new pods should start shortly :)
    ₳ЯṬ
    @chisNaN
    thanks MAN
    but same here my describe quota return 10 pods while get pods return no resources
    still
    DavidCamelo
    @DavidCamelo
    hi @erulabs I have all my pods down
    image.png
    Seandon Mooy
    @erulabs
    Hello everyone - things should be repaired - ideally your pods have been online off and on during all this hassle. I've verified everyone who posted here last night has their pods up and running.. Been a crazy 24 hours - our master servers came under very heavy attack.. We're looking at our options for providing a reliable hosting service - unfortunately the free tier is becoming extremely hard to defend. We'll follow up on this in the next couple days... Rest assured paying users: things should only get more reliable going forward :)
    I'm going to take a very very very needed nap, but please feel free to ping here if you're having an issues, and I'd be happy to take a look!
    The resource quota issue should also be resolved now - this was also due to the master servers being under tremendous load. Thanks for your patience everyone!
    Seandon Mooy
    @erulabs
    Just finished a double check and things look really healthy. It's possible some pods are in a ContainerCreating or Unknown state - you should be able to delete those pods and they'll get re-created right away.
    Carl J. Mosca
    @carljmosca
    can we get an explanation of what happened when convenient? I apologize if I missed it
    Dan Pastusek
    @PastuDan
    @carljmosca Seandon was the one who fixed, but I can try and explain: AWS unexpectedly terminated two of our masters (out of 3 total) for our USW1 cluster. When we spun them back up, the certs we generated were invalid (our mistake was that Kube requires masters to all have the exact same certificates, rather than just ones signed by the same CA). This caused the ectd database on the 2 new masters to fall behind, and therefore everyone's resource quota (stored on etcd) was out of sync, so the workers to not be able to schedule any new pods (also exacerbated by the cert issue when talking to the masters). Fixing those certs and rebooting the masters fixed the cluster and allowed pods to be scheduled again.
    Carl J. Mosca
    @carljmosca
    Thank you Dan
    Dan Pastusek
    @PastuDan
    @carljmosca no problem
    apologies for the outage!
    Carl J. Mosca
    @carljmosca
    I like to think we’re all always learning. As opposed to not. 😶
    YenRaven
    @YenRaven
    Humm, my pods are crashing on startup right when they begin downloading their first mod on startup?
    ₳ЯṬ
    @chisNaN
    I would give the advise to do kubectl describe quota and relaunch the web app if you dont see anything because it seems the web ui is not updating well...
    ₳ЯṬ
    @chisNaN
    one thing i dont understand is why describe quota return 0 ingresses while i have 1 active
    progressbarteam
    @progressbarteam
    Seems like ip address by which I was reaching my services exposed by NodePort changed. How can I check the new one?
    Seandon Mooy
    @erulabs
    You should be able to “kubectl get pods -o wide” to get the ip address where the pod lives
    weve been thinking about this - are considering allowing dedicated IPs for paying customers
    progressbarteam
    @progressbarteam
    Yes, i am familiar with “kubectl get pods -o wide” and those addresses, but recently i was able to reach all of my services by single ip 13.56.80.18
    progressbarteam
    @progressbarteam
    It probably was node ip, wasn't it?
    MO
    @8Mobius8

    Hey there! Ran into a tiny problem with my free tier 'repo' setup on Kubesail.com: Looks like the 'repo' functionality actually builds the image in the kubesail's private registry but my pods can't start due to a Unauthorized error when trying to get the image to run. Is the private registry accessible for free tier accounts?

    The build seems to show as passing on the UI, however when I look at the pods I see a pull image error:

     failed to resolve reference "registry.kubesail.com/....
     failed to authorize: failed to fetch anonymous token: unexpected status: 401 Unauthorized
    Seandon Mooy
    @erulabs
    Interesting - yes free tier repo buillder should work - and yes it uses our private registry... but it should also create a Secret and attach the ImagePullSecret to the deployment.
    Ill take a look - perhaps the Deployment had the imagePullSecret removed or our system failed to patch it on correctly...
    4 replies
    we need some better docs around that feature!
    @progressbarteam yes that was a node IP - unfortunatly this weekend we had a bit of an issue which moved lots of pods onto new nodes. Were working on a better solution - and you should be able to use
    ext-a.kubesail.io Instead of any IP
    but thats a bit of a beta system - let me know if it works for you :)
    YenRaven
    @YenRaven
    @erulabs is there possibly something with the new setup you can think of that would cause my pods to crash while trying to download files? I can connect to the pod for a few seconds through the kubesail site but then they stop responding and crash right after logging out the init of the first download.
    Seandon Mooy
    @erulabs
    errr... no... I suppose possibly outbound networking on a particular host could be having an issue, but I'd suspect Id see some alarms on my side for that... It's very much more likely that something is erroring inside the program.
    Typically an issue with our infrastructure would prevent pods from starting at all - they wouldnt effect what goes on once they launch
    YenRaven
    @YenRaven
    Hey it works now! In fact, I think I got everything working now! Well, everything so far. I can actually join a game on my app now with mods and all!
    I do have one question though. I can only join on the Node port, but to be able to allow people to join from the public list, it has to be available on 34197 UDP. Is there something I can do to correct that?
    YenRaven
    @YenRaven
    nvm, I was able to change the NodePort to 34197 and direct connect via that but it still didn't work via public list connect. Guess I got something else wrong.
    Seandon Mooy
    @erulabs
    Hrmmm - i know factorio has some complicated logic for displaying on the shared games list
    im actually tempted to patch it - factorio is brilliant but i dont think the developers do much devops :P
    i believe the problem is that factorio tries to determine its public IP - since its inside a container it doesnt guess right :(
    YenRaven
    @YenRaven
    Looks like factorio headless has a --port and --bind flag to hopefully handle this but I tried setting both and they hit me with this log. Lol " Setting both --port and --bind with port is dumb and you should not do it."
    Seandon Mooy
    @erulabs
    Lol!
    but “bind” is a bad name - we dont want to bind a specific address - we just want to advertise a specific address
    it should have an —advertise-address, but if i recall last time i looked it did not :(
    progressbarteam
    @progressbarteam
    ext-a.kubesail.io works great! Thanks!
    YenRaven
    @YenRaven
    I'm actually running into some UX issues with the site and managing environment variables. On FF once I create a variable, I cannot delete it. Clicking the 'x' to remove the variable once seems to shift the page slightly. A second time deletes the variable from the list, but when I click save, it comes right back with it's old value.