Odd NFS issues
Hi!
I have set up some Debian boxes, running Sarge, Kernel 2.6.11, patched
with GRSecurity 2.1.3 set in medium mode (although I am fairly sure
that's not related)
I basically mounting some filesystems off another Debian box (also
2.6.11 but NO Grsec) and I am experiencing some wierd NFS issues whereby
some (but not all) mounts just disappear. Let me illustrate with some
"before" and "after" outputs from df: (excuse bad formatting please)
Before:
store:/share/area1
583342464 18441856 564900608 4% /local/area1/files
store:/share/area2
583342464 18441856 564900608 4% /local/area2/files
store:/share/area3
583342464 18441856 564900608 4% /local/area3/files
store:/share/admin/cron/area1
20428 367552 1672576 19%
/local/area1/var/spool/cron/crontabs
store:/share/admin/auth/area1
20428 367552 1672576 19%
/local/area1/etc/passwords
store:/share/admin/tmp/area1
20428 367552 1672576 19% /local/area1/tmp
store:/share/admin/cron/area2
20428 367552 1672576 19%
/local/area2/var/spool/cron/crontabs
store:/share/admin/auth/area2
20428 367552 1672576 19%
/local/area2/etc/passwords
store:/share/admin/tmp/area2
20428 367552 1672576 19% /local/area2/tmp
store:/share/admin/cron/area3
20428 367552 1672576 19%
/local/area3/var/spool/cron/crontabs
store:/share/admin/auth/area3 20428 367552 1672576 19%
/local/area3/etc/passwords
store:/share/admin/tmp/area3 20428 367552 1672576 19% /local/area3/tmp
After:
store:/share/admin/cron/area1
- - - -
/local/area1/var/spool/cron/crontabs
store:/share/admin/auth/area1
- - - -
/local/area1/etc/passwords
store:/share/admin/tmp/area1
- - - - /local/area1/tmp
store:/share/area1
583342464 18441792 564900672 4% /local/area1/files
store:/share/admin/cron/area2
- - - -
/local/area2/var/spool/cron/crontabs
store:/share/admin/auth/area2
- - - -
/local/area2/etc/passwords
store:/share/admin/tmp/area2
- - - - /local/area2/tmp
store:/share/area2
583342464 18441792 564900672 4% /local/area2/files
store:/share/admin/cron/area3
- - - -
/local/area3/var/spool/cron/crontabs
store:/share/admin/auth/area3
- - - -
/local/area3/etc/passwords
store:/share/admin/tmp/area3
- - - - /local/area3/tmp
store:/share/area3
583342464 18441792 564900672 4% /local/area3/files
The mounts have myteriously turned into hyphons and I cannot access them
on the command line - issuing a mount -o remount does not fix the issue,
but unmounting the affected mounts and issuing mount -a -t nfs does
fix it every time.
Note the filesystems above which did not vanish are always the ones that
do not vanish, and the ones which did are the ones that always do. All
of the filesystems are mounted with the same options, and all are shared
with the same options on the server side.
Interestingly, I have several clients all mounting the same stuff, and
when they turn into hyphons, they do on ALL clients (indicating a
server issue to me), but even more interestingly, if you just unmount
and remount the affected mounts on one client, it fixes ALL clients!
The options the mounts are shared with are (rw,sync,no_root_squash) on
the server end, and it has RPCNFSDCOUNT set to 64 (although I only have
6 clients)
The filesystems are mounted on the clients with these fstab options:
tcp,hard,nointr,timeo=600,retrans=2,rw,nolock,bg
When they go, nothing is logged either on the server or the client,
and they seem to be going at the rate of every couple of hours at the
moment, without heavy accessing.
I am stumped and google has not been my friend. Does anyone at all
recognise these symptoms or have any advice as to what I can look
at to fix it?
Many thanks
GW
--
_ _ _a' /( <. # Gaz Wilson, aka DragonLord
/_/ _ _/ / ` / _ ~~ _}\ \( _ ) # E-Mail: dragon@nightwolf.org.uk
/ \ /_'/_/ /_, / /_||/|/ \(,_(,)' # Info: http://www.dragons.org.uk
Catch me with the Dragons. ._>, _>, # Chat: http://www.redclaw.org.uk/
Reply to: