Best guess: The

pre-upgrade

command was introduced in Longhorn

v1.5.0

for https://github.com/longhorn/longhorn/issues/5131. The

v1.4.3

longhorn-manager image does NOT have it, but the

v1.5.5

image DOES.

eweber@laptop:~/longhorn-manager> docker run --rm longhornio/longhorn-manager:v1.4.3 longhorn-manager pre-upgrade
panic: unrecognized command: pre-upgrade

eweber@laptop:~/longhorn-manager> docker run --rm longhornio/longhorn-manager:v1.5.5 longhorn-manager pre-upgrade
NAME:
   longhorn-manager pre-upgrade - 

USAGE:
   longhorn-manager pre-upgrade [command options] [arguments...]

OPTIONS:
   --kube-config value  Specify path to kube config (optional)
   --namespace value    Specify Longhorn namespace [$POD_NAMESPACE]
   
time="2024-07-26T19:01:46Z" level=fatal msg="Critical error: Required flag \"namespace\" not set" func=main.main file="main.go:77"

By default the

v1.5.5

chart uses the

longhornio/longhorn-manager:v1.5.5

container for the pre-upgrade hook. It seems you are using a values.yaml, etc. that references an older version of longhorn-manager without the command.

It now works perfectly