lookupdotdot failing again

[A Dog]

On Sat, Jun 19, 2021 at 10:56:52AM PDT, Matthew Dillon wrote:
> Yah, keep monitoring it.  I kinda suspect what is happening is that one or
> more processes on the clients are winding up CD'd into NFS directories that
> then get renamed or deleted/replaced, or something like that.  But it could
> easily also have been a hash collision in the file handle calculation the
> server makes.
> 
> -Matt

This happened again the other night.  The entire dmesg buffer consisted
of iterations of this message:

lookupdotdot failed 2 dvp 0xfffff800e9597e80

I'm not aware of a concomitant client-side failure, but my guess is that
there must have been one.  I'm also highly doubtful that this is due to
directory renames/deletions; this isn't my first NFS rodeo and I know
what can happen when you do that, and it was happening before in
directories that are only ever used by one client and/or have had the
same name and inode number for years.

My fear is that NFS access is going to start failing again left right
and center before too long; it happened after about a month of uptime
last time.

Is there any additional information I could provide that could help in
diagnosing the problem?  Unfortunately this is a production file server
so the degree to which I can mess with it while it's running (and the
amount of downtime I can get away with) is limited...


-- 
A Dog