Skip to content

tcarmet/k8s-connection-reset

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 

Repository files navigation

k8s-connection-reset

Sharing code to reproduce an issue with Kubernetes

Requisites

To install the following code you'll need

  • A kubernetes cluster.
  • kubectl.

Install steps to reproduce the issue

Deploying with skaffold

If you have skaffold installed.

$ skaffold run
# or
$ skaffold dev

Deploying with kubectl

$ kubectl apply -f kube

⚠️ Important note ⚠️

Some of the clients pod must be scheduled on a different node (it can be just one different node) to reproduce the issue.

We identified that when everything was scheduled on only one node the problem was harder to reproduce.

Identifying the issue

Check on Stackdriver logging the log from the server and client. on the search bar you can set text:'reset by peer' and you'll see the following errors:

# Error reported by the server
E  [info] 11#11: *21885 client 10.20.6.249 closed keepalive connection (104: Connection reset by peer)
# Error reported by the client
E  curl: (56) Recv failure: Connection reset by peer

Both line should follow each other. If the client and the server are not reporting at the same time a Connection reset by peer you can ignore the event as it might be something else.

Troubleshooting

If the clients seem stuck feel free to restart the pods like the following kubectl delete pod -l app=client

Links

Relates to:

About

Sharing code to reproduce an issue with Kubernetes

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published