submariner-io / submariner

Networking component for interconnecting Pods and Services across Kubernetes clusters.
https://submariner.io
Apache License 2.0
2.42k stars 190 forks source link

Submariner connectivity lost for 15-20 minutes #2572

Closed dirajag1 closed 1 year ago

dirajag1 commented 1 year ago

We have our OCP cluster connected to 2 switches in an MLAG pair . We upgraded one of the switches and observed that when the switch went for a reboot , submariner connectivity went down and was degraded for 15-20 minutes . The connectivity was restored after that without any manual intervention. Also observed several of the submariner pods had restarted couple of times. Attaching subctl gather logs for reference
must-gather.zip

skitt commented 1 year ago

Which version of Submariner are you running? Your subctl gather output says 0.12.0 — is that accurate?

dirajag1 commented 1 year ago

Yes, 0.12 is the submariner version

skitt commented 1 year ago

Are you staying on 0.12.0 for any specific reason?

dirajag1 commented 1 year ago

We are still on OCP 4.10 , and moving to newer versions of submariner would mean we also upgrade our OCP version . We will go there , but it may be few months down the line

sridhargaddam commented 1 year ago

@dirajag1 It looks like you are using OCP 4.10 with OVN-K CNI. Please note that Submariner is supported with OVNK CNI starting from OCP 4.11+ version as documented here - https://submariner.io/getting-started/architecture/networkplugin-syncer/ovn-kubernetes/

You have two choices.

  1. Change the CNI from OVN-K to OpenShift SDN while continuing to use OCP 4.10
  2. Move to a newer version of OCP and a more recent version of Submariner.

Personally, I would suggest option 2.

dfarrell07 commented 1 year ago

Closing because incompatible versions