CONTP-100757：使用Cisco 9337C -拧 秒共享交换机的MetroCluster IP中的SSD和NVM驱动器报告了高延迟

问题描述

High latency might be observed on NVMe drives of NS224 storage shelves in MetroCluster IP configurations that use Cisco Nexus 9336C-FX2 shared switch configurations. EMS log:
[ClusterA-01: disk_latency_monitor: shm.ssd.threshold.ioLatency:notice]: SSD e5a.03.0.11 has exceeded the expected block latency in the current timeframe with an average latency of 1828 us and an average utilization of 8 percent. The next highest SSD latency: 1684 us. Disk e5a.03.0.11 Shelf 3 Bay 11 [NETAPP X4011WBORA3T8NTF NA51] S/N [22106S123456] UID [0014EE83:0B31A280:00000000:00000000:00000000:00000000:00000000:00000000:00000000:00000000]
Switch log might report:
%ACLQOS-SLOT1-2-ACLQOS_UNEXPECTED_PFC_FRAMES: Ethernet1/17 received 4 unexpected PFC frames for COS 3 %ACLQOS-SLOT1-2-ACLQOS_UNEXPECTED_PFC_FRAMES: Ethernet1/18 received 2 unexpected PFC frames for COS 3