NFS server in Slackware 14.2 slow

Mig21 · 07-15-2016, 10:07 PM

Hi

I upgraded my server to Slackware 14.2 and on the NFS client (Linux Mint) I have a permanent NFS mount. And it seems to mount ok. But when I open some directories in Nemo (the default Mint/Cinnamon file manager) it takes a very long time (like a minute).

I think for the same reason it takes me a long time to log into the graphical interface on the workstation.

Strangely it seems to work fine in a terminal inside the graphical interface. I can run a find on the directory and it takes 3 seconds.

Any sugestions what could be the cause? The only change I see on the server is it seems in 13.37 NFS was version 3 and in 14.2 it's version 4.

Here's a bunch of commands I ran with no helpful results:

Quote:

# Server:

cat /etc/exports
/data 10.0.0.*(rw,sync,no_root_squash,subtree_check)

exportfs -ra
<no output>
echo $?
0

showmount localhost
clnt_create: RPC: Program not registered

dmesg | tail -8
[ 22.200841] r8169 0000:02:00.0 eth0: link up
[ 22.200861] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[ 22.940369] NFSD: the nfsdcld client tracking upcall will be removed in 3.10. Please transition to using nfsdcltrack.
[ 22.940379] NFSD: starting 90-second grace period (net ffffffff8251c780)
[ 143.242307] NFSD: Unable to end grace period: -110
[ 209.043098] NFSD: Unable to create client record on stable storage: -110
[41876.129843] NFSD: Unable to create client record on stable storage: -110
[41911.206369] NFSD: Unable to create client record on stable storage: -110

dmesg | grep -i nfs
[ 0.289435] RPC: Registered tcp NFSv4.1 backchannel transport module.
[ 0.327588] NFS: Registering the id_resolver key type
[ 0.327854] Installing knfsd (copyright (C) 1996 okir@monad.swb.de).
[ 22.940369] NFSD: the nfsdcld client tracking upcall will be removed in 3.10. Please transition to using nfsdcltrack.
[ 22.940379] NFSD: starting 90-second grace period (net ffffffff8251c780)
[ 143.242307] NFSD: Unable to end grace period: -110
[ 209.043098] NFSD: Unable to create client record on stable storage: -110
[41876.129843] NFSD: Unable to create client record on stable storage: -110
[41911.206369] NFSD: Unable to create client record on stable storage: -110

ps ax | grep nfs
50 ? S< 0:00 [nfsiod]
1027 ? S< 0:00 [nfsd4_callbacks]
1030 ? S 0:00 [nfsd]
1031 ? S 0:00 [nfsd]
1032 ? S 0:00 [nfsd]
1033 ? S 0:00 [nfsd]
1034 ? S 0:00 [nfsd]
1035 ? S 0:00 [nfsd]
1036 ? S 0:00 [nfsd]
1037 ? S 0:00 [nfsd]
1631 pts/0 S+ 0:00 grep nfs

ps ax | grep rpc
27 ? S< 0:00 [rpciod]
1023 ? Ss 0:00 /usr/sbin/rpc.rquotad
1039 ? Ss 0:00 /usr/sbin/rpc.mountd
1155 ? Ss 0:00 /sbin/rpcbind -l
1160 ? Ss 0:00 /sbin/rpc.statd
1635 pts/0 S+ 0:00 grep rpc

cat /etc/nfsmount.conf | grep Defaultvers
Defaultvers=3 # I changed it to 4, made no difference

/usr/sbin/rpc.nfsd 8 -d
rpc.nfsd: knfsd is currently up

rpcinfo -p localhost
program vers proto port service
100000 4 tcp 111 portmapper
100000 3 tcp 111 portmapper
100000 2 tcp 111 portmapper
100000 4 udp 111 portmapper
100000 3 udp 111 portmapper
100000 2 udp 111 portmapper
100024 1 udp 35338 status
100024 1 tcp 44811 status

# Workstation:

showmount -e 10.0.0.3
clnt_create: RPC: Program not registered

mount | grep nfs
nfsd on /proc/fs/nfsd type nfsd (rw,relatime)
10.0.0.3:/data on /home/andrew/data type nfs4 (rw,relatime,vers=4.0,rsize=262144,wsize=262144,namlen=255,hard,proto=tcp,port=0,timeo=600,retrans=2 ,sec=sys,clientaddr=10.0.0.4,local_lock=none,addr=10.0.0.3)

There's some nonsense about IPv6, showmount not working, nfsd not showing in rpcinfo, but I don't know what to do now.

Please help! This is driving me crazy!

perbh · 07-15-2016, 10:44 PM

I have given up on nfs (since a while back - like the entirety of 14.1) - now using sshfs instead and no problems.
Yes - I had the same problems as you and in addition, when using 'apps' (for lack of a better word) to access files more than 2 levels below the mountpoint - it would fail with 'file not found'. cp, cat, ls etc worked fine, but not vi (as short examples).
In the end I gave up trying to figure out what was happening ...

Mig21 · 07-15-2016, 10:46 PM

Hm, I can give that a shot, thanks for the suggestion. Can you share the appropriate line from fstab? I've only ever used sshfs manually after logging in.

allend · 07-15-2016, 11:06 PM

Quote:

rpcinfo -p localhost
program vers proto port service
100000 4 tcp 111 portmapper
100000 3 tcp 111 portmapper
100000 2 tcp 111 portmapper
100000 4 udp 111 portmapper
100000 3 udp 111 portmapper
100000 2 udp 111 portmapper
100024 1 udp 35338 status
100024 1 tcp 44811 status

That does not look right to me. I see

Quote:

root@host:~# rpcinfo -p localhost
program vers proto port service
100000 4 tcp 111 portmapper
100000 3 tcp 111 portmapper
100000 2 tcp 111 portmapper
100000 4 udp 111 portmapper
100000 3 udp 111 portmapper
100000 2 udp 111 portmapper
100024 1 udp 32765 status
100024 1 tcp 32765 status
100011 1 udp 32769 rquotad
100011 2 udp 32769 rquotad
100011 1 tcp 32769 rquotad
100011 2 tcp 32769 rquotad
100003 3 tcp 2049 nfs
100003 4 tcp 2049 nfs
100003 3 udp 2049 nfs
100003 4 udp 2049 nfs
100021 1 udp 32768 nlockmgr
100021 3 udp 32768 nlockmgr
100021 4 udp 32768 nlockmgr
100021 1 tcp 32768 nlockmgr
100021 3 tcp 32768 nlockmgr
100021 4 tcp 32768 nlockmgr
100005 1 udp 32767 mountd
100005 1 tcp 32767 mountd
100005 2 udp 32767 mountd
100005 2 tcp 32767 mountd
100005 3 udp 32767 mountd
100005 3 tcp 32767 mountd

Is /etc/rc.d/rc.nfsd set as executable?
This may also help. http://docs.slackware.com/howtos:net...home_nfs_howto

BTW - Slackware still uses NFS version 3

Mig21 · 07-15-2016, 11:22 PM

Quote:

BTW - Slackware still uses NFS version 3

That was it, thanks! I added nfsvers=3 to the fstab and now it seems to work. What a relief.

Weird thing is when this all started I had nfsvers=3 (or something like it) in the fstab on my clients and they said something like "unavailable nfs version selected". Maybe I had it as nfsvers=v3 before, I can't remember.

Also the "RPC: Registered tcp NFSv4.1 backchannel transport module." was a confusing message

The output of rpcinfo also changed:

Quote:

rpcinfo -p localhost
program vers proto port service
100000 4 tcp 111 portmapper
100000 3 tcp 111 portmapper
100000 2 tcp 111 portmapper
100000 4 udp 111 portmapper
100000 3 udp 111 portmapper
100000 2 udp 111 portmapper
100024 1 udp 53254 status
100024 1 tcp 60236 status
100011 1 udp 773 rquotad
100011 2 udp 773 rquotad
100011 1 tcp 776 rquotad
100011 2 tcp 776 rquotad
100003 3 tcp 2049 nfs
100003 4 tcp 2049 nfs
100003 3 udp 2049 nfs
100003 4 udp 2049 nfs
100021 1 udp 59124 nlockmgr
100021 3 udp 59124 nlockmgr
100021 4 udp 59124 nlockmgr
100021 1 tcp 40866 nlockmgr
100021 3 tcp 40866 nlockmgr
100021 4 tcp 40866 nlockmgr
100005 1 udp 34763 mountd
100005 1 tcp 42891 mountd
100005 2 udp 44096 mountd
100005 2 tcp 50538 mountd
100005 3 udp 39812 mountd
100005 3 tcp 57893 mountd

Weird, could it be that it changes once a client connect correctly?

I hope this wasn't a fluke. I'll reboot a few more times

linuxbawks · 02-13-2018, 12:28 PM

Good grief, all these ports in the wiki need to be opened up for NFSv3 to function correctly. Having said that you can still mount with NFSv4 as in:

Code:

sudo mount -t nfs -o vers=4 <server_ip>:/src_path /dest_path

However operations such as read/write are slow. I've had to open the rpc ports and am using NFSv3. It seems it's also impossible to boot a rootfs using NFSv4 as the kernel refuses to mount the rootfs. Again I had to resort to NFSv3 with all those rpc ports opened up.