-
-
Notifications
You must be signed in to change notification settings - Fork 83
Description
I just upgraded to latest version via the helm chart, and the updated didnt go through.
I have a mulitnode K3s Cluster and gave the pdfding-pod (replica 1) a RWO volume before hand.
During the update the new pod was scheduled on another node, and ofc it failed to be created because it couldnt attach the volume again.
This is due to the default update strategy for the deployment (as its no where configured).
strategy:
rollingUpdate:
maxSurge: 25%
maxUnavailable: 25%
type: RollingUpdateI suspect this hasnt been noticed yet because tests run in a single node cluster where different pods can mount the same pvc on the same node, or it was just luck and old and new pod got scheduled on the same host.
Anyway, it would be great if the Chart had the option to set the depoloyment strategy to "recreate" DOCS
For anyone coming across this issue: I set replica: 0 and then back to 1 and that solves it.