Does message draining only work when the cluster is scaled?

[lavifb]

Sometimes we may want to restart all the pods, like if we are upgrading to a new version of kubernetes and need to move to new nodes. Is there a way to maintain availability while restarting the pods one by one?

[gaohoward]

It should safe to restarting the pods without losing data.

[lavifb]

but are the messages drained to the other nodes? Or are they just saved in the pv for when the restarted node comes back?

[brusdev]

yes the messages saved in the pv comes back when the node is restarted

[gaohoward]

make sure you enabled messageMigration (if persistenceEnabled is true then it by default enabled).