High CPU/Disk utilization on Snapmirror source node from Deswizzlers

Applies to

• ONTAP 9
• ONTAP (7-mode)
• Active IQ Unified Manager

Issue

• High CPU utilization observed due to data processing work
• Latency is higher than expected after a DP SnapMirror relationship is broken or a volume move finishes
• qos statistics workload resource cpu show shows that this is due to _WAFL_SCAN
• wafltop show -v io shows scanners consuming most resources.
• wafl scan status shows many deswizzlers running
• Alert messages stating Node Performance Limit - System Node Utilization from Active IQ Unified Manager (AIQUM)
• High resource consumption on single node

Example: A single deswizzle is running for volume1:

filer> priv set advanced
filer*> wafl scan status

Volume volume1:
 Scan id                   Type of scan     progress
  226328    container block reclamation     block 1871 of 24095 (fbn 11220)
  226329             volume deswizzling     snap 157, inode 2144843 of 2172681. level 1 of normal files. 
  Totals: Normal files: L1:10/4881 L2:0/11336 L3:0/5723 L4:0/0 Inode file: L0:0/0 L1:0/0 L2:0/0 L3:0/0 L4:0/0