0

I have a server that's to be used for imaging here, and the client machines boot off NFS roots. Or, rather, a single NFS root.

Here's the problem: when only one client is connected, the system runs as expected, but if there are multiple clients, there's a reasonably good chance that I'll get a stale file handle error.

I'm at my wit's end, here. What can I do to prevent errors on a heavily shared NFS mount?

3
  • Multiple clients can not share the same root directory. They need their own individual /var (and /dev?). Where are the stale filehandle occurring?
    – Kusalananda
    Commented May 15, 2017 at 19:24
  • They have their own /dev and their own /var/{log,lock,tmp,run}. Commented May 15, 2017 at 19:53
  • And where is the stale filehandle usually occurring?
    – Kusalananda
    Commented May 15, 2017 at 20:17

1 Answer 1

-1

Apparently, adding /var/log to the list of things mounted tmpfs was enough to get it working. Leaving this here for the future generations.

In my use case, I don't need logging unless the machine is still running, so a tmpfs is fine. If you need logging, there are ways to fix this issue that won't die on reboot.

Big takeaway is that if you're having issues on machines with NFS roots, such as applications crashing at random, check what /var/log is up to.

2
  • Wouldn't call mounting /var/log on a temporary file system "fixing it"...
    – Kusalananda
    Commented May 15, 2017 at 20:56
  • For my use case, at least, it's more than sufficient :) Logging is only useful to me while the machine is running, but if I ever need to have persistent logging, it's an easy fix to make - the big takeaway is that /var/log is the root of my problems. Commented May 16, 2017 at 12:11

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .