[OmniOS-discuss] Fwd: NFS server unresponsive

[Schweiss, Chip]

---------- Forwarded message ----------
From: Schweiss, Chip <chip at innovates.com>
Date: Wed, Feb 1, 2017 at 8:58 PM
Subject: NFS server unresponsive
To: zfs <zfs at lists.illumos.org>


I have a storage pool with 100+ NFS clients.   It is part of an HA cluster.
  No matter which host I put the pool on NFS comes to a halt for all pools.
  Everything else is fine.  Just no NFS.

Most of the clients are NFSv4.  I've rebooted the system and the problem
repeats itself in just a few seconds of being responsive after the pool
come online.

I suspected NFS resources, and tried doubling most everything:

# sharectl get nfs
servers=1024
lockd_listen_backlog=512
lockd_servers=4096
lockd_retransmit_timeout=5
grace_period=90
server_versmin=3
server_versmax=4
client_versmin=3
client_versmax=4
server_delegation=on
nfsmapid_domain=
max_connections=-1
protocol=ALL
listen_backlog=64
device=
mountd_listen_backlog=64
mountd_max_threads=16

Didn't change anything.

Nothing in logs.   Pool is fine for everything but NFS.

nfssvrtop show no I/O about 99% of the time, but periodically shows a
couple small I/O.


Possibly a client knocking over, but I cannot isolate the issue tighter
than the entire pool.

What else should I be looking at?

Any help greatly appreciated.

-Chip
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://omniosce.org/ml-archive/attachments/20170201/e10e9173/attachment.html>