ssh_exchange_identification: read: Connection reset by peer

[eugenewww]

$ git push deis-staging master
ssh_exchange_identification: read: Connection reset by peer
fatal: Could not read from remote repository.

Workflow was installed 2 years ago on GKE. How can I debug it and resolve the issue? This issue happened several days ago.

[eugenewww]

Looks like this issue is related to teamhephy/builder. Should I open the issue there and close it here?

[Cryptophobia]

Looks like an ssh connectivity problem. Can you check the SSH key that you are using is valid in pushing to the git deis-builder remote?

Also, it could be that the IP or domain has not changed for the deis-builder git remote:

git remote get-url deis-staging

Another thing to check is to see if the deis remote url is correct and the name of the app is still correct in the above output.

Also check if the deis-builder is up and running:

kubectl get pods -n deis | grep -i "deis-builder"

[eugenewww]

Hi @Cryptophobia. Key is correct. Remote is correct. I can't deploy changes for any app. No one from my team can do it too.

$ kubectl get pods -n deis | grep -i "deis-builder"
deis-builder-1364651073-tmkzz                            0/1     CrashLoopBackOff   2107       548d

[eugenewww]

$ kubectl --namespace=deis logs deis-builder-1364651073-tmkzz 2019/11/17 19:29:17 Error creating storage driver (googleapi: got HTTP response code 400 with body: )

[Cryptophobia]

This seems to be the same error some users are experiencing with deis-builder on off-cluster Google Storage objectstore. Some of them have migrated to S3 in order to mitigate the issue while it is being investigated by the Google Cloud support team.

Are you able to join our Slack channel and see what the latest is on this issue? https://slack.teamhephy.com

[eugenewww]

@Cryptophobia thanks.

[Cryptophobia]

Haven't heard any more information related to this issue so I will go ahead and close it for now. Join our Slack or report another issue if this persists.