Hi all,
I'm debugging some likely buggy network device driver. Let's say the
driver erroneously looses some packets (under certain conditions). One
of important tests is being able to mount and use NFS share (as a
client). I'm currently using kernel 4.14.221 and NFS v3 in UDP mode.
While testing, I've noticed some strange NFS behaviour. If some NFS
request (file read) obtains no response (let's suppose due to
connectivity issue), such request will be resent periodically (in hope
to eventually succeed). This is expected. Now I observe, that after I
unmount the NFS share (successfully), the failed requests still keep
being resent, apparently infinitely (Confirmed by e.g. Wireshark),
surviving also ifdown and ifup. So basically, I've found no way to get
rid of them other than reboot.
Is this a known behaviour? Is it by design? Does it make any sense?
Thank you,
Regards,
Nikolai