High Latency Within Stack

[jpeterprarch]

Has anyone encountered an issue with high latency within a MS350 stack?  We recently had an event where suddenly one day extremely high latency (800-1000ms) was occurring between the stack's management IP (that also serves as the gateway for remote devices) and some (not all) hosts connected to the stack.  The latency only occurred for those who were remotely connected to VPN, or in branch offices. Very bizarre.  Restarting the entire stack resolved the issue... ...for now.

[PhilipDAth]

Did you measure this ping?  If so, that may not be accurate.  ICMP processing is a low-priority task.  The actual routing is done in silicon.  You could ping a ping of 1000ms but a packet forwarding latency of 500 microseconds.

One of the things that can consume a lot of CPU time (but not affect packet forwarding) are spanning tree issues and processing lots of multicast joins.

Next time grab a packet capture for a post-incident review.