useful-one-liners

screen

• Start a named session screen -S name
• list of screens: screen -ls
• re-attacn to a screen: screen -r session-name
• kill a screen: ctrl + a and then k

Docker

• Build and Run:

docker build -t my_docker .
docker run -p 9000:9000 my_docker

• Connecting to the container and geting a shell: docker container run -it: start the container interactively (i for interactive and t for pseudo tty)

docker container exec -it: run additional command in existing container

docker exec -it <mycontainer> /bin/bash

• Delete all containers in docker:

docker rm $(docker ps -a -q)

• some useful arguments:

  • --name to name a container
  • --detach run in the detached mode
  • --env (-e) pass in an environment variable docker container stop [list of names] to stop containers

• Inspecting containers:

docker container top CONTAINER: list all running processes of a container

docker container inspect: displaying details of one or more container

docker container stats: displays perfromance stats for all containers in a real-time stream

Kuberneters

• Get list of pods:

kubectl -n namespace get pods

• To see more details

kubectl describe po/name

• To see the logs: kubectl logs po/podname

• To see the running logs:

kubectl logs -f po/podname

• Connecting to the pod

kubectl exec -it pod-id /bin/bash

• After connecting the pod, You can do port-forwarding to see the tensor board: Open another tab and run tensorboard:

Tensor board —logdir=path_to_logdir

and then portforward:

kubectl port-forward pod-id 6006:6006

• Copy to and from a pod:
  • Copy /tmp/foo local file to /tmp/bar in a remote pod in namespace

kubectl cp /tmp/foo <some-namespace>/<some-pod>:/tmp/bar

• Copy /tmp/foo from a remote pod to /tmp/bar locally:

kubectl cp <some-namespace>/<some-pod>:/tmp/foo /tmp/bar

s3

sync an s3 bucket where only keys mathig specific patterns are included:

aws s3 sync --sse aws:kms --sse-kms-key-id keyid s3://s3bukcet/keys/ mylocalpath --exclude '*' --include *.csv' --include '*.tiff'