Fabric MetroCluster high write latency
Applies to
- Fabric MetroCluster
- ONTAP 9
- Brocade FC switches
- ATTO 7500N/7600N fibrebridge
Issue
- Write latency is high across all SVMs and volumes.
porterrshow
insupportshow
output from the backend switches shows no errors.- EMS logs show repeated SCSI command aborted errors all indicating the same bridge:
Mon May 22 11:25:54 +0200 [cluster01_node01: slifc_intrd:scsi.cmd.abortedByHost:error]: Disk device switch001:7.126L39: Command aborted by host adapter: HA status 0x4: cdb 0x88:000000014c718748:00000008.
Mon May 22 11:26:47 +0200 [cluster01_node01: slifc_intrd:scsi.cmd.abortedByHost:error]: Disk device switch001:7.126L31: Command aborted by host adapter: HA status 0x4: cdb 0x9a:000000014c79ae00:0001:0200.
Mon May 22 11:27:57 +0200 [cluster01_node01: slifc_intrd:scsi.cmd.abortedByHost:error]: Disk device switch001:7.126L36: Command aborted by host adapter: HA status 0x4: cdb 0x28:d9746598:0018.
- Latency returns to a normal level when the switch port connected to the affected bridge is disabled.