E.g. on my Kubernetes 1.10.9 cluster the DaemonSet got created with OnDelete update policy for whatever reason, which gave me a hard time finding out why weren't pods getting re-deployed after helm upgrade, nevertheless there was the checksum/config annotation update - until I found https://github.com/helm/helm/issues/2735#issuecomment-442860510.
It's even worth to consider enforcing the strategy to be type: RollingUpdate by default - so to align with expectations from checksum/config annotation (which is hard coded ATM).
A user may want to update it.
E.g. on my Kubernetes 1.10.9 cluster the DaemonSet got created with
OnDelete
update policy for whatever reason, which gave me a hard time finding out why weren't pods getting re-deployed afterhelm upgrade
, nevertheless there was thechecksum/config
annotation update - until I found https://github.com/helm/helm/issues/2735#issuecomment-442860510.It's even worth to consider enforcing the strategy to be
type: RollingUpdate
by default - so to align with expectations fromchecksum/config
annotation (which is hard coded ATM).