Arch Linux

skaloo

Member

Registered: 2026-03-12

Posts: 3

nfsrahead crashes in the logs - exploration, (possible) solution

Hello,

I recently installed a new server, on which I added both nfs and cifs auto-mounts through systemd.
(So far i had servers with only nfs or only cifs).
The whole stuff worked fine, but then I noticed those errors in the logs (everytime a mount was auto-done):

kernel: CIFS: Attempting to mount XXXXXXXXXXXXXXXXX
systemd-coredump[47778]: Process 47777 (nfsrahead) of user 0 terminated abnormally with signal 6/ABRT, processing...
systemd[1]: Started Process Core Dump (PID 47778/UID 0).
systemd-coredump[47779]: [?] Process 47777 (nfsrahead) of user 0 dumped core.
                                                         Stack trace of thread 47777:
                                                         #0  0x00007f8e160baa2c n/a (libc.so.6 + 0x98a2c)
                                                         #1  0x00007f8e160601a0 raise (libc.so.6 + 0x3e1a0)
                                                         #2  0x00007f8e160475fe abort (libc.so.6 + 0x255fe)
                                                         #3  0x00007f8e16048697 n/a (libc.so.6 + 0x26697)
                                                         #4  0x00007f8e160c4cbc n/a (libc.so.6 + 0xa2cbc)
                                                         #5  0x00007f8e160c4cd4 n/a (libc.so.6 + 0xa2cd4)
                                                         #6  0x000055b06889e35a n/a (/usr/bin/nfsrahead + 0x235a)
                                                         #7  0x00007f8e160496c1 n/a (libc.so.6 + 0x276c1)
                                                         #8  0x00007f8e160497f9 __libc_start_main (libc.so.6 + 0x277f9)
                                                         #9  0x000055b06889e5e5 n/a (/usr/bin/nfsrahead + 0x25e5)
                                                         ELF object binary architecture: AMD x86-64
(udev-worker)[47776]: cifs-32: Process '/usr/bin/nfsrahead cifs-32' terminated by signal ABRT.

Notice this is a cifs mount and it calls 'nfsrahead' - which doesn't seem appropriate.

I found out that the nfs-utils package contains /usr/lib/udev/rules.d/99-nfs.rules, which contains a single rule

SUBSYSTEM=="bdi", ACTION=="add", PROGRAM="/usr/bin/nfsrahead %k", ATTR{read_ahead_kb}="%c"

So arguably, I'm not super-familiar with that part of the linux system, but it doesn't seem right to me to start nfsrahead for just any block device ?

It seems sensible to me to restrict that to the nfs ones, so I added

ENV{ID_FS_TYPE}=="nfs"

to the rule, and the logs are now clean.

But as I said, I'm not super-familiar with that part of the system and I would like to know if that's a proper change to do. And if so, shouldn't that already be done in the package file ?

Thanks,

srahman5317

Member

Registered: 2021-11-19

Posts: 4

Re: nfsrahead crashes in the logs - exploration, (possible) solution

I am getting the same error. Also just a point to note. According to systemd changes the correct way to implement rules would look like:

ACTION!="remove", SUBSYSTEM=="bdi", PROGRAM="/usr/bin/nfsrahead %k", ATTR{read_ahead_kb}="%c"

Just wanted to add that making the above change doesn't actually solve the problem. Changing ENV as suggested by OP does.

Last edited by srahman5317 (2026-03-12 17:18:40)