Hi team, we observed there's a lot of network comm...
# troubleshooting
s
Hi team, we observed there's a lot of network communication btw task managers for stateful streaming pipeline, especially with large states. For stateless streaming service, do we expect a lot network communication between each task managers as well?
r
might happen there's a lot of shuffle for those jobs
s
state shuffle?
Yeah I guess the network communication is because of state shuffle. So I guess it shouldn't happen on stateless service?
r
yup
( Also i saw a post from you regarding the FLINK SQL infra set up on top of k8s ) Can. you check my post if you have faced similar issues ?
s
can you share the link?