0

i have major issues on one of my 14.04 systems. On another one everything works totaly fine (this 2 Ubuntu machines are running on a hyper-v host, so they have exactly the same networking). The cifs server is a QNAP NAS device.

I receive periodically this message: [ 8436.827930] CIFS VFS: Server 192.168.2.210 has not responded in 120 seconds. Reconnecting... Also sometimes I have seen this errors: [ 9352.315504] CIFS VFS: Send error in Close = -11

and access to the cifs is extremely slow. I have mounted it manually via: mount -t cifs //192.168.2.210/aaa/bbb/ccc /mnt/msnas-repos -o username=admin,password=aabbcc,domain=local,dir_mode=0777,file_mode=0777

here some information from debug output: [ 8310.492638] /build/buildd/linux-lts-utopic-3.16.0/fs/cifs/sess.c: serverOS=QTS [ 8310.492640] /build/buildd/linux-lts-utopic-3.16.0/fs/cifs/sess.c: serverNOS=Samba 3.6.25 [ 8310.492643] /build/buildd/linux-lts-utopic-3.16.0/fs/cifs/sess.c: serverDomain=NETWORK

Here some timings: - working system: /mnt/msnas-repos# time mkdir -p test2/test/test

real    0m0.030s
user    0m0.000s
sys     0m0.001s

- non-working system: time mkdir -p test2/test/test

real 9m31.120s user 0m0.000s sys 0m0.002s

As you can see the same command requires more then 9 minutes, and my assumption is 99,9% of this 9 minutes is just cifs sleeping around.

Any idea what I can check? As the both systems are installed the same it must be some configuration difference but as I am no Ubuntu specialist I have no glue where to start with. Also all systems patch level are exactly same. The NAS utilization is extremely low and when i am writing a 1G testfile from the working system the speed is ok: 

/mnt/msnas-repos# dd if=/dev/zero of=tempfile bs=1M count=1024 conv=fdatasync,notrunc
1024+0 Records in
1024+0 Records out
1073741824 Bytes (1.1 GB) copied, 16.548 s, 64.9 MB/s

Any idea would be great. thx.

michL
  • 101
  • ok. i found out the issue only appears when MRTG is running.And because it will be so slow all the mrtg processes get stapled and make it even worse every 5 minutes. Will check with them whats the problem. – michL Nov 27 '16 at 14:31
  • Gitlab published some information a while ago and they concluded that with any disk latency CIFS basically is unusable. What kind of disks are backing CIFS? – Kristopher Ives Nov 27 '16 at 14:50
  • @KristopherIves thank you for your note, there is not really any disk latency as it is on the same network (even same room) and the disks are very fast. i have measured a disk write speed about 180 MB/s on this disks. – michL Nov 27 '16 at 15:05
  • @michL for further reference, it's nice to post your output in English, not German/Dutch. –  Nov 27 '16 at 15:28

0 Answers0