CONTAP-242986: Timeouts and latency on nfs lookups with FPOLICY
Issue
- After upgrading ONTAP from version 9.9.1P16 to 9.13.1P8, Intermittent performance issue that causes commands like
- Observing latency on nfs lookups:
Index Procedure Calls Min SRT Max SRT Avg SRT Sum SRT 1 GETATTR 23 0.000387 0.000519 0.000445 0.010242 3 LOOKUP 6856 0.001472 34.995091 0.007256 49.745074 4 ACCESS 19 0.000388 0.000587 0.000459 0.008728 16 READDIR 137 0.000555 0.002521 0.001063 0.145611 17 READDIRPLUS 40 0.000428 0.009099 0.001039 0.041568 18 FSSTAT 1 0.000588 0.000588 0.000588 0.000588
================================================- NFS operations give up and return jukebox error to the client, this can be observed in packet traces :
- [NFS 22 V3 LOOKUP Reply (Call In 141723) Error: NFS3ERR_JUKEBOX]
- and in SKTRACE :
- [[0:0] NBLADE_Chk: n0x7b9510: src/Protocols/Nfs/NfsFPolicyProcess.cpp:1636 rt=3288]
- [[0:0] NBLADE_Chk: n0x62a615: src/Protocols/Nfs/Nfs3/Nfs3ProcRename.cpp:249 rt=3288]
- [[14:0] NBLADE_Chk: n0x7b9510: src/Protocols/Nfs/NfsFPolicyProcess.cpp:1636 rt=3288]
- [[14:0] NBLADE_Chk: n0x617331: src/Protocols/Nfs/Nfs3/ Nfs3ProcLookup.cpp:274 rt=3288]
- Disabling FPOLICY stops the issue