We have a 4-node cluster. 3 of them are also serving disks as OSD's for their Ceph cluster, the 4th is not.
The 4th has also slightly faster CPU.
The 3 servers are Dell R620 with a Dell raid controller, the other one is a HP ProLiant something with HP raid card probably then.
On the 3 nodes that are similar, I see constant Recv-Q in netstat for the corosync process.
This is also visible on some underlying UDP processes that run in the containers.
What could be the source of this Recv-Q? What I do need to troubleshoot?
The 4th has also slightly faster CPU.
The 3 servers are Dell R620 with a Dell raid controller, the other one is a HP ProLiant something with HP raid card probably then.
On the 3 nodes that are similar, I see constant Recv-Q in netstat for the corosync process.
This is also visible on some underlying UDP processes that run in the containers.
What could be the source of this Recv-Q? What I do need to troubleshoot?