Deploying helloworld dockerfile example not working on digitalocean cluster #2438
Comments
|
@loungerider what's the output of |
|
core@deis-1 ~ $ fleetctl status feline-duckling_v2.cmd.1.service |
|
@loungerider if you're inside one of the boxes, you need to make sure you forward your key to the SSH agent. If you ssh'd manually into the box via something like |
|
Got it, logged into deis-2 and ran the command. core@deis-2 ~ $ fleetctl status feline-duckling_v2.cmd.1.service Nov 09 04:05:35 deis-2 systemd[1]: Failed to start feline-duckling_v2.cmd.1. |
|
core@deis-2 ~ $ etcdctl get /deis/registry/host |
|
@loungerider what version of CoreOS is this on? Could you also please post |
|
@bacongobbler see below. Yes, probably related. core@deis-2 ~ $ cat /etc/os-release core@deis-2 ~ $ docker --version |
|
@bacongobbler moby/moby#8887 - definitely a known issue. Looks like updating the Deis registry service will probably make the most sense. |
|
I just ran into this -- turns out that DigitalOcean is still on 493.0.0. I opened up a ticket with them on this and will update when I get more information |
|
response from DigitalOcean: |
|
@bacongobbler I just opened a ticket as well, hopefully they push 494.0.0 with the docker revert to the alpha channel soon. It's obvious the 1.3.1 implementation was a fail. What's the Deis team thoughts on changing the registry service? Can I use the coreos stable channel in the interim? |
|
@loungerider in the interim, you can update to v494 via http://docs.deis.io/en/latest/managing_deis/upgrading-deis/#upgrading-coreos. We want to get to the beta channel but we require a minimum of CoreOS v471 for ceph fs kernel fixes. Docker fixes are coming upstream |
|
Thanks, I'm good until the channel is updated. |
|
@bacongobbler A couple of things I noticed after upgrading to 494.0.0. Not a show stopper. I still get this error on the every git push deis master from the helloworld example or from the app I'm trying to deploy. When running deis start platform I get the following errors: I can ctrl c and run deis start platform again and all services start eventually and are active. |
|
@loungerider Those are errors that bubble up from fleet. That typically means that fleet is getting overloaded on one of your machines. If you |
|
DigitalOcean has updated to 494 now, so we should be good again. We'll keep tracking the docker 1.3.1 issue in #2395. If you have any other issues that are not related to DigitalOcean provisioning v493.0.0, please feel free to open another issue. Thanks! |
|
I just provisioned a new cluster on DigitalOcean with Deis v1.0.0. I deployed example-dockerfile-python and everything seems happy: http://casual-ziggurat.mattboersma.com/ |
|
👍 |
I believe a change made over the last few days has caused an issue with dockerfile deployment with Deis. The same deploy worked on Thursday night. I've been testing out Deis, removing and recreating the cluster. This morning I cloned a new deis repo and followed the cluster install as documented. The cluster comes up, all services are running, I can register and git push deis master "works", the docker image build runs, pushes to the private registry and then tries to launch. The launch fails, it takes a long time to respond and ends with an error.
The coreos cluster instances show the container trying to activate while toggling between start-pre and auto-restart. Noticed this when deploying my app. Removed the digitalocean droplets recreated the cluster and I was able to reproduce the issue. I'm using the Deis digitalocean cloud-config and the helloworld dockerfile example.
Let me know if you need more log info.
deisctl journal controller
The text was updated successfully, but these errors were encountered: