[helm loki/promtail] update of DaemonSet takes a long with a lot of nodes

[stefanandres]

Is your feature request related to a problem? Please describe.
When deploying a change for the promtail DaemonSet, the configured default is to only rollover one pod at a time.
This may take 1 minute per pod. When you have a Cluster of more than a few Nodes, this will take an immensive amount of time for simples changes.

Current default:

  updateStrategy:
    rollingUpdate:
      maxUnavailable: 1
    type: RollingUpdate

Describe the solution you'd like
Since having a slow rollover does not give you any more availablility, we may configure to allow more parallel rollover.

Additional context

I'd suggest to change the MaxUnavailble settings in the UpdateStrategy.

Now we can also discuss a sane default for the helm chart.

1. The option should be configureable
2. What default should be use here? If we continue to use 1 updates are slow, but misconfiguration of the pod would mean that only one logging Pod would be broken and not all of them. If we use something like 100% options are fast, but when the Pod is misconfigured all logs would be broken.

Let me know what you think of it and I can whip up some PR ;)

[cyriltovena]

I think 25% is a sane default. And yes we would love a contribution to improve this.

Thank you :)