...making Linux just a little more fun!

<-- 2c Tips | TAG Index | 1 | 2 | 3 | 4 | 5 | 6 | 7 | Knowledge Base

The Answer Gang

By Jim Dennis, Jason Creighton, Chris G, Karl-Heinz, and... (meet the Gang) ... the Editors of Linux Gazette... and You!



(?) Not a simple NFS question (2.6.11.11)

From Justin Piszcz

Answered By: Thomas Adam

I have a Debian Etch (testing) machine running 2.6.11.11.
Using FTP, I can easily achieve 10MB/s+ (megabytes per second) to or from the box. The main HDD in the box is a 400GB, 7200RPM, ATA/100 HDD attached to an ATA/133 Promise Controller.
I believe the problem began when I upgraded the box to 2.6.11.
Problem Description:
Sending files to the box via NFS is slow, 1MB/s for a few seconds, nothing, then 2-3MB/s then back to nothing.
Receiving or (pulling) files from the box achieves a sustained transfer rate of 10MB/s+ (megabytes per second).
Boxname = p4box.
p4box (share = /home/myfiles)
If I mount p4box onto another box:
# mount p4box:/home/myfiles /mnt
# cp 1gbfile.txt /mnt
The above process will take forever, stall, lag and etc.
If I mount p4box onto another box:
# mount p4box:/home/myfiles /mnt
# cp /mnt/1gbfile.txt .
I get normal NFS speeds on a 100mbps network.

The problem also occurs in the following scenario:
On p4box (box with the problem):
I:
# mount remote_machine:/disk2 /mnt
# cp /mnt/1gig.txt .
This also does average 1MB/s, but not consistent, it peaks and goes slower and then nothing at all for a while.
If I do the same and copy something to the NFS share (outbound) then it runs at 10MB/s (megabytes per second).

The issue is inbound NFS is VERY slow.
On the affected box:
# Kernel Interface table
Iface   MTU Met   RX-OK RX-ERR RX-DRP RX-OVR   TX-OK TX-ERR TX-DRP TX-OVR Flg
eth0   1500 0   1283895      0      0      0  636062      0      0      0 BMRU
lo    16436 0      1040      0      0      0    1040      0      0      0 LRU
There is clearly no network issue.
REMEMBER: FTP works at full speed in both directions.
# rpcinfo -p l1
    program vers proto   port
     100000    2   tcp    111  portmapper
     100000    2   udp    111  portmapper
     100021    1   udp  32768  nlockmgr
     100021    3   udp  32768  nlockmgr
     100021    4   udp  32768  nlockmgr
     100024    1   udp    831  status
     100024    1   tcp    834  status
     100003    2   udp   2049  nfs
     100003    3   udp   2049  nfs
     100005    1   udp    996  mountd
     100005    1   tcp    999  mountd
     100005    2   udp    996  mountd
     100005    2   tcp    999  mountd
     100005    3   udp    996  mountd
     100005    3   tcp    999  mountd
I've tried rebooting the box, disconnecting/remounting/etc, nothing seems to solve this issue.
I've searched Google and the Web but I have not found anything pertinent to this situation.
Does anyone have any idea what I can try next?
Thanks!
Justin.
(!) [Thomas Adam] You haven't told us how you mount your NFS share -- i.e. what options you have contained in /etc/fstab. You can control the data rates with:
wsize=
rsize=
which are values you can specify (usually in KiB) so that you don't throttle anything.

(?) Pretty standard options:

l2:/d3           /l2/d3         nfs     rw,hard,intr,nfsvers=3D3 0 0

These options are on three boxes, each with its own NFS share and I only have the problem on 1 machine.


(?) The problem is solved.

The switch I was using went bad.


This page edited and maintained by the Editors of Linux Gazette
HTML script maintained by Heather Stern of Starshine Technical Services, http://www.starshine.org/


Each TAG thread Copyright © its authors, 2005

Published in issue 118 of Linux Gazette September 2005

<-- 2c Tips | TAG Index | 1 | 2 | 3 | 4 | 5 | 6 | 7 | Knowledge Base
Tux