VSX cluster down after adding new VS

[Wolfgang]

Today we built a new virtual system with new interfaces. We used an interface which was not in use and added some VLANs. Everything looks fine.

But after installing the policy for the first time we lost connectivity. More then a little bit panic in the team. We‘re in production and everything was lost. After getting console access we saw one member in state „ready“ and the other in state „down“. Bringing down the new VS via cpstop on both nodes in the context of this VS and after cpstop/cpstart on the down member we are back again and online.

Digging the logs we could see some error messages regarding the new interface, no communication possible between both nodes via this interface. This was correct because the switchports were disabled at this moment.

Yes, these interfaces couldn’t communicate, but why goes down the whole VSX cluster and not only the new VS does not start? These new interface is used only in the new VS, nothing else.

Is this the normal behaviour? I can‘t remember how we added new VS last time with interfaces up or not.

[Chris_Atkinson]

PRJ-30284 ?

ClusterXL

In VSX Load Sharing (VSLS) environment, a disconnected bond LS interface impacts all VS's at the member regardless that the interface is connected to a specific VS.

[Wolfgang]

We are running VSX R80.40 take 156, no VSLS, new added interface is no bond only single one on both nodes.