When wrongly configuring s3-integrator Postgresql charm will get stuck in error state and needs to be redeployed #661
Description
Steps to reproduce
Deploy postgresql charm
Follow backup and restore tutorial
I forgot to set up tls-ca-chain config that was required in this case
And now postgresql charm gets stuck in error state not being able to get out of it wihout hacking inside charm hooks or redeploying it
Expected behavior
Not to get stuck in error state if the user makes a configuration mistake but to get into blocked state
Actual behavior
# juju status
[..]
postgresql/0 active idle 0 192.168.30.106 5432/tcp
postgresql/1* error idle 1 192.168.30.138 5432/tcp hook failed: "s3-parameters-relation-changed"
postgresql/2 active idle 2 192.168.30.152 5432/tcpYou can NOT get out of this state even
- By re-running the hook
- By updating the config
- By updating the config and re-running the hook
- By removing the relation
- By removing the relation and re-running the hook
Versions
Operating system: Ubuntu 22.04
Juju: 3.5.4
Charm revision: 14/stable rev468
Log output
Did not collect it at the moment
Additional context
You could hackily get out of this situation wihout redeploying by skipping the hook and then removing the relation
- On a terminal "A"
juju debug-hooks postgresql/1
# Now you are on a tmux session waiting for a hook to get triggered- On a terminal "B"
juju resolved postgresql/1
# Trigget the hook to re-run- On terminal "A"
# tmux session is waiting for you to manually execute the reaction to the s3-parameters-relation-changed hook
exit 0
# to make juju belive the hook was run okay
exit
# to exit the juju-debug sessionNow the hook has been skipped and you can remove the relation
Update the config
And re-add the relation with the proper configuration