High latency spikes on a MetroCluster IP system due to bursts of NVRAM traffic
Applies to
- ONTAP 9
- MetroCluster IP
Issue
- Intermittent latency spikes reported by customer
- EMS reporting congestion:
Wed Mar 01 13:01:14 +0100 [NodeA1: mccip_mirror_congestion_mgr_p: mcc.network.congestion:notice]: Network congestion detected. Action taken: Increased ic_timeout to 1200 msec.
- PerfArchives show high NVRAM latency
- Back-end switches showing discards for NVRAM traffic (Q5)
