Hello again! Is there any guidance on how to handl...
# flyte-v2-support
b
Hello again! Is there any guidance on how to handle workload disruptions with node pool updates when using flyte 2 binary? I did a node pool update today that rerolled the node flyte 2 was running on and a couple of workloads got aborted as a result
f
really can you share how long was the update for
this is not expected, also the flyte-core is in progress that will provide more replicas to you can do a rolling deployment
🙏 1
🦜 1
cc @glamorous-carpet-83516 this should not happen right, controller should keep retrying
g
it should not happen. do you remember when did it get aborted. is it immediately after you restart flyte pod?
b
Yea - it happened after flyte pod was restarted, I think the outage was probably somewhere around 1-3 mins from node spin down to a new one spinning up
g
@faint-pilot-24699 mind taking a look at this
f
Will take a look
@brash-nest-94989 I tried to reproduce this. Flyte will recreate the pod when a pod is evicted. However, it will abort an action when the CR is deleted(CR's DeletionTimestamp is marked), and I'm thinking that might be the reason for your issue. Would you able to check whether that happened when you reroll your node pool?