Grab container ID using bash sorcery to pass to container_kill_start kill task.

[und1sk0]

• @anandkumarpatel
• @kaushikanurag

Stepping through the changed command:

sudo docker ps | grep -v '^CONTAINER' | awk '{print $1,$2 }' | grep -e 'registry.runnable.com/runnable/navi:reconfigure-newrelic' -v -e : |awk -F' ' '{print $1}'

sudo docker ps

• self explanitory

grep -v '^CONTAINER'

• strip the header line

awk '{print $1, $2}'

• not only do we only care about the first to columns, we are trying to match untagged containers by looking for the absence of a ":" (see next command)

grep -e {{ container_id }} -v -e :

• match the container_id (registry.runnable.com/<some_service>:<some_branch>) or any untagged container

awk -F' ' '{print $1}'

• print the container's hash

[anandkumarpatel]

have you tried docker inspect <containerID> on the lost container? that might have better info on something we can grab to tell if it is lost container for that branch. also look into docker ps --format < you can get anything from the output of docker inspect will help reduce the logic in your grep statement

[und1sk0]

We could but it seems like now we are piling on logic.

Here's a mini-flow, then:

CONTAINER_ID=sudo docker ps | grep {{ container_id }} | awk '{print $1}'
if [ "" != "${CONTAINER_ID}" ] ; then
echo $CONTAINER_ID
else
for ID in sudo docker ps | grep -v '^CONTAINER' | awk '{print $1}' ; do
sudo docker inspect ${ID} | grep -q {{ container_id }}
if [ ${?} ] -eq 0 ; then
OUR_ID=sudo docker inspect ${ID} | awk -F" '{print $4}'
# and in an abundance of caution
CONTAINER_ID="${OUR_ID} ${CONTAINER_ID}"
fi
done
echo $CONTAINER_ID
fi

[und1sk0]

At this point might as well write a shell script, upload it, and run it...

[anandkumarpatel]

is this needed?

[und1sk0]

Sorry, was for debugging.

[anandkumarpatel]

remove comment, documentation is in the name

[anandkumarpatel]

one small nit, but good to go

[anandkumarpatel]

up to your discretion, should this be in /ops/* path?