Justin Garrison

You can get all child processes with ppid option. Something like this should work

ps h -o pid --ppid 1 | xargs strace

This would obviously require xargs in your support container but that shouldn’t be an issue.

You have to do it manually right now buy finding the node the container is running on an then ssh to the host and follow the same steps. I’m not sure if kubernetes will add similar features but you still have access to the containers on the nodes.

Just thought I’d pass this along that maybe it’ll be helpful working with ingress. https://medium.com/@rothgar/exposing-services-using-ingress-with-on-prem-kubernetes-clusters-f413d87b6d34#.mlacwjm45

You’ve probably already found a solution but I thought I’d share.

I have used sysdig since ~2013. It’s a great tool but it only allows for passive monitoring (can’t send command to running process) and if you want to use tracers you have to have the source code of the application running in a container.

A great option for monitoring and finding root cause for outages but not an option for…

Correct but I didn’t want to make assumptions about someone’s cluster and I wanted to show how you can find the node a pod is running on.

An unfortunate situation with using Medium is I don’t have options to change the font. :( Otherwise I would have changed it a long time ago.

Thanks for asking. You made me realize that the traefik RC had changed (I fixed the post) and if you look at the spec for the deployment you’ll see it specifies which hostport it wants to use

The microbot pod exposes port 80 only on the pods IP which is part of a subnet and not directly accessible unless you have access to the network being used by the pods. The traefik pod exposes port 80 on the pod and also maps that port to 80 on each node port. This was if you go to any node on port 80 you’ll be sent to port 80 of the pod (which then…

If the ingress service is properly exposing the NodePort on 80 you shouldn’t have this problem. Jump in kubernetes.slack.com if you still have problems and many people can help.

Thanks for the article. Just 2 tips to hopefully make some things easier for you.

1. You don’t have to cat the yaml and pipe to kubectl. You can just do kubectl create -f file.yaml
2. You don’t need to do that ugly sed substitute on the file. Instead use NAMESPACE=foo envsubst file.yaml | kubectl create -f -