How to identify NFSv4 drop package count by autosupportdata ?

brucelinapp — Thu, 05 Jun 2025 05:34:17 GMT

Could someone can help me to explanation autosupport log for nfsstat -D ?

The key point is to identify drop package count by NFS share with NFSv4.

Any document or website is OK. Thanks

ASUP sample :

Server nfs V4: (6277500544 calls, 18327100049 ops)

null compound badproc2 access close commit

69 6277500475 0 0% 440552271 2% 452541668 2% 0 0%

create delegpurge delegret getattr getfh link

14668930 0% 0 0% 11245700 0% 5168175870 28% 739379883 4% 0 0%

lock lockt locku lookup lookupp nverify

169491704 1% 171 0% 160307672 1% 309924661 2% 0 0% 0 0%

open openattr open_confirm open_downgrade putfh putpubfh

455048548 2% 0 0% 192361 0% 7036 0% 6276499803 34% 0 0%

putrootfh read readdir readlink remove rename

241 0% 674990945 4% 120672735 1% 2 0% 59035063 0% 17962455 0%

renew restorefh savefh secinfo setattr setclntid

14550415 0% 487672010 3% 487679879 3% 0 0% 173921565 1% 78 0%

setclntid_cfm verify write rlsowner

78 0% 0 0% 2088165902 11% 4412403 0%

Read request stats (version 4)

0-511 512-1023 1K-2047 2K-4095 4K-8191 8K-16383 16K-32767 32K-65535 64K-131071 > 131071

3064546 11566 25696082 1044457 456588689 14469234 114862405 959586 58294380 0

Write request stats (version 4)

0-511 512-1023 1K-2047 2K-4095 4K-8191 8K-16383 16K-32767 32K-65535 64K-131071 > 131071

697563561 97164422 58018325 157818925 946813592 45135022 7944163 5216526 72491367 0

Misaligned Read request stats

BIN-0 BIN-1 BIN-2 BIN-3 BIN-4 BIN-5 BIN-6 BIN-7

1781791256 0 0 0 0 0 0 0

Misaligned Write request stats

BIN-0 BIN-1 BIN-2 BIN-3 BIN-4 BIN-5 BIN-6 BIN-7

2685726830 19711 117493 17649 5584346 17827 59907 17863

NFS V2 non-blocking request statistics:

null getattr setattr root lookup readlink read

126128 100% 0 0% 0 0% 0 0% 0 0% 0 0% 0 0%

wrcache write create remove rename link symlink

0 0% 0 0% 0 0% 0 0% 0 0% 0 0% 0 0%

mkdir rmdir readdir statfs

0 0% 0 0% 0 0% 0 0%

NFS V3 non-blocking request statistics:

null getattr setattr lookup access readlink read

141 100% 1768 0% 0 0% 13 0% 851 0% 0 0% 0 0%

write create mkdir symlink mknod remove rmdir

3 0% 0 0% 0 0% 0 0% 0 0% 696 0% 0 0%

rename link readdir readdir+ fsstat fsinfo pathconf

0 0% 0 0% 0 0% 0 0% 14 0% 0 0% 0 0%

NFS reply cache statistics:

TCP:

InProg hits Misses Cache hits False hits

0 2512051253 0 0

UDP:

In progress Misses Cache hits False hits

0 0 0 0

nfs reply cache size=381600, hash size=6151

flows alloc'd=267, max flows=3816

flows used=77, flows free=190

reserve entries=95, nflow LRU=0, grow LRU=162, opinfo releases=0

entry alloc fail=0, reply alloc fail=0, flow alloc fail=0, connection drops=0

Connection drops because of in progress hits:

v3 conn dropped=0

v4 conn dropped, no reconnect=0

Re: How to identify NFSv4 drop package count by autosupportdata ?

bbjholcomb — Wed, 18 Jun 2014 19:08:50 GMT

What NIC card do you have in your system(s) (use sysconfig -ac to get the model number and the card chipset)?

What version of Ontap are you running?

You should check the card slot the NIC is in to make sure you are getting maximum out of the PCIe slot. Some controllers have PCIe x2 and x4 slots for example.

Are you running Jumbo frames from end to end?

The chipset is very important, I believe NetApp only shows drop packets for Chelsio, nothing for Intel or QLogic. Which NIC card you have is very important.

We experienced drop packets recently, it took a while to get the information, we have over 100 7mode controllers (NFS only) most had a fair amount had drop packets. Keep in mind dropping packets on Ethernet connections is common, the amount compared to total packets, if you are you dropping 1 out of 50 million packets, most people consider that reasonable.

If you do a packet capture on the client side (limit the amount of data), use WireShark, go into expert mode, click on Analyze then exprtinfo.

If a NFS client is sending or receiving a fair amount of data and a packet is dropped it can cause the filer to retransmit all of the packets if you set net.ip4.tcp_sack=1 (selective acknowledgement) on the NFS client, then only the dropped packet will be retransmitted. This made a big differences for us on Oracle logwr messages.

I would also look at the NFS mount options for the client machine, you cut down the number of setattr and getattr.

I plan to write up how we gathered our information on dropped packets and the differences in the 10G NIC cards, once I do I will post on this community web page.

I hope this useful.

topic Re: How to identify NFSv4 drop package count by autosupportdata ? in Network and Storage Protocols

How to identify NFSv4 drop package count by autosupportdata ?

Re: How to identify NFSv4 drop package count by autosupportdata ?