CONTAP-45637:HDDでI/Oキューが高いため読み取りレイテンシが発生
問題
- Similar to bug 1444991, high queued i/o is seen on a specific drive or set of drives.
- Example:
[cluster1::*> statistics start -object disk -counter disk_busy|io_pending|io_queued|total_transfers -sample-id disk_queued Statistics collection is being started for sample-id: disk_queued] - Example: - The Disk:4.1.18 & 4.1.30 , both are serving almost equal no. of IOPS, however the IOs queued on Disk 4.1.18 is high and the Disk is 100% busy. - Look for one or few such disks in the system. [cluster1::*> statistics show -sample-id disk_queued Object: disk Instance: 4.1.18 Start-time: 2/13/2023 10:05:11 End-time: 2/13/2023 10:05:17 Elapsed-time: 6s Scope: cluster1-01 Number of Constituents: 1 (complete_aggregation) Counter Value -------------------------------- -------------------------------- disk_busy 100% <-- io_pending 7.99 io_queued 600.08 <-- total_transfers 378