-
Notifications
You must be signed in to change notification settings - Fork 1.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Ability to configure DIND bridge network #1652
Comments
@FleetAdmiralButter Hey! This does seem like a valid feature request. |
I think this is happening to me too. |
@ctr26 Hey! Thanks for chiming in- please feel free to submit a pull request 😄 |
I tried to get around this by using |
This doesn't actually change the MTU for me
|
As a work around we turned off docker in the RunnerDeployment since we ran into the same problem. Good feature request though |
I ended up disabling docker too, which causes its own problems but it's fine |
@ctr26 Currently,
Which secret are you referring, K8s secret or GitHub Actions secret? I thought it works with both though. |
GitHub actions secrets were failing, I'll check again though. |
@jlees-Gss Thanks. But where do that 5 minutes come from? 🤔 I thought |
@ctr26 Thanks. Interesting! It would be great if you could share related logs or errors. If it's coming from workflow jobs, it might be an issue in |
@ctr26 can you try with the dind image please |
That was a typo above, I did try it with the DIND image and the correct volume mounts, the MTU even changed but I still had the same problem. |
So according to the log I think it first has to timeout looking for the daemon.
Unless I've missed something in the RunnerDeployment that needs changing.
EDIT: I found the DISABLE_WAIT_FOR_DOCKER field resolves start up time |
@jlees-Gss Thanks! Another solution to your problem is to "unset" |
@mumoshu Thanks for the advice We then verified by kubectl exec -it into the pod and was able to use dockerd ps We're still in the process of verifying our dockerfile workflows are working as should bee I'll share the code just now Create configmap of new daemon.json
Attach configmap to runnerdeployment as volume
Just for your information this problem was occurring in AWS for us |
@jlees-Gss I feel like you've got a really close workaround here. I'm in a similar boat, although docker doesn't seem to be live inside my runner container. Would you be willing to share the whole yaml? I'm in the same boat you are injecting daemon.json via configmap, and I'm ending up with the repeated message @mumoshu I'm going through the code, but I'm afraid I don't see the part of the code in the runner that STARTS docker... do you have any info on this? |
It appears to be possible to use the This works (on AWS) for me without needing to disable waiting for docker or utilizing
|
@dputnamfr your solution works for me on Azure, AKS cluster as well |
Is your feature request related to a problem? Please describe.
The default Docker bridge network, 172.17.0.0/16, overlaps with our VPC network. This breaks network connectivity within the DIND container.
Describe the solution you'd like
Similar to the dockerMTU parameter, could there be a parameter that could be used to override the default Docker bridge network?
Describe alternatives you've considered
Alternatively, could the controller implement a way to specify a different DOCKER_HOST other than localhost? I'm able to get a separate DIND deployment running by manually setting the bridge network, but there doesn't seem a way to make ARC connect to it.
The text was updated successfully, but these errors were encountered: