Major issues with nfsv4

[Konstantin Belousov]

On Sun, Dec 13, 2020 at 05:08:48PM +0000, Rick Macklem wrote:
> J David wrote:
> [stuff snipped]
> >The most common trace (91 times, or over 1/4 of all observed) is:
> >
> >__mtx_lock_sleep+0xf8 nfscl_nodeleg+0x207 nfs_lookup+0x314
> >VOP_LOOKUP_APV+0x75 null_lookup+0x98 VOP_LOOKUP_APV+0x75 >lookup+0x451
> >namei+0x414 kern_statat+0x72 freebsd11_stat+0x30 amd64_syscall+0x387
> >fast_syscall_common+0xf8
> This is just waiting on the mutex that protects the open/lock/delegation
> state.
> 
> Now, is this just caused by heavy load and 130000 opens, or is there
> something in nullfs that results in more state handling than needed?
> [more stuff snipped]
Nullfs with -o nocache (default for NFS mounts) should not cache vnodes.
So it is more likely a local load that has 130k files open.  Of course,
it is the OP who can answer the question.