Improve Debugging Capabilities For Helm Application Installs
It is very difficult to debug when a helm application install fails as we only give generic messages in the UI.
We should at least do the following improvements:
-
Actually log errors somewhere so that we can view these later for debugging if necessary -
Add set -x
to install pod commands -
Remove all > /dev/null
from install commands since this makes it harder to see what happened -
Don't delete the failed installation pod until you retry. So we should do a "delete if exists" before running the install again. Same thing will be necessary for the config map except I guess this is (find or create). Then we should update the messaging to users to say "Installation failed, check the pod logs for install-<app-name>
"
Edited by Dylan Griffith