[Gluster-users] AFR not working ?

[a_pirania at poczta.onet.pl]

Hi!
We have some problems with glusterfs. Configuration: 2x clients and 2x
servers. On both clients we run this loop:

#!/bin/bash


for ((i = 0; i < 100000000; i++)) {

KAT=$RANDOM
mkdir /mnt/glusterfs/$KAT
for ((j=0; j< $RANDOM; j++)) {
PLIK=$RANDOM.$RANDOM
echo "PLIK: $KAT/$PLIK "
dd if=/dev/urandom of=/mnt/glusterfs/$KAT/$PLIK bs=$RANDOM count=1
dd if=/mnt/glusterfs/$KAT/$PLIK of=/dev/null
rm -f /mnt/glusterfs/$KAT/$PLIK
}
rm -rf /mnt/glusterfs/$KAT
}

Both servers are running, and everything looks ok. When I stoped one server
lopp are going and everything looks ok. When, afftr 1-2 minutes I started
this stopped earlier server, on one or both clients dd can't write/read
file:

PLIK: 32586/21510.20986
dd: opening `/mnt/glusterfs/32586/21510.20986': No such file or directory
dd: opening `/mnt/glusterfs/32586/21510.20986': No such file or directory
PLIK: 32586/8965.12208
dd: opening `/mnt/glusterfs/32586/8965.12208': No such file or directory
dd: opening `/mnt/glusterfs/32586/8965.12208': No such file or directory
PLIK: 32586/6812.30905
dd: opening `/mnt/glusterfs/32586/6812.30905': No such file or directory
dd: opening `/mnt/glusterfs/32586/6812.30905': No such file or directory
PLIK: 86/14676.13776
1+0 records in
1+0 records out
20064 bytes (20 kB) copied, 0.00489 seconds, 4.1 MB/s
39+1 records in
39+1 records out
20064 bytes (20 kB) copied, 0.002158 seconds, 9.3 MB/s
PLIK: 86/4774.7934
1+0 records in
1+0 records out
14855 bytes (15 kB) copied, 0.004141 seconds, 3.6 MB/s
29+1 records in
29+1 records out
14855 bytes (15 kB) copied, 0.001101 seconds, 13.5 MB/s
PLIK: 86/20883.24292
1+0 records in
1+0 records out
9717 bytes (9.7 kB) copied, 0.002439 seconds, 4.0 MB/s
18+1 records in
18+1 records out
9717 bytes (9.7 kB) copied, 0.0007 seconds, 13.9 MB/s
PLIK: 86/13210.19257
1+0 records in
1+0 records out
31787 bytes (32 kB) copied, 0.007434 seconds, 4.3 MB/s
62+1 records in
62+1 records out


Afetr few secunds everything its fine. We think that glusterfs, when secund
server connecting after shutdown, transfers I/O to this newly connected
server and generates errors. How to fix it ?


client:

volume client1
type protocol/client
option transport-type tcp/client
option remote-host 10.1.100.130
option remote-port 6996
option remote-subvolume posix1
end-volume

volume client2
 type protocol/client
 option transport-type tcp/client
 option remote-host 10.1.100.131
 option remote-port 6996
 option remote-subvolume posix2
end-volume

volume afr
type cluster/afr
subvolumes client1 client2
end-volume

volume rh
 type performance/read-ahead
option page-size  100KB
option page-count 3
subvolumes afr
end-volume

volume wh
type performance/write-behind
option aggregate-size 1MB
option flush-behind on
subvolumes rh
end-volume


server:

volume posix2
type storage/posix
option directory /var/storage/glusterfs
end-volume

volume posix-locks
type features/posix-locks
option mandatory on
subvolumes posix2
end-volume

volume io-thr
type performance/io-threads
option thread-count 2
option cache-size 64MB
subvolumes posix-locks
end-volume

volume server
type protocol/server
option transport-type tcp/server
option listen-port 6996
subvolumes io-thr
option auth.ip.posix2.allow 10.*.*.*
end-volume



version:
glusterfs 1.3.12 built on Dec 11 2008 13:41:06
Repository revision: glusterfs--mainline--2.5--patch-797
Copyright (c) 2006, 2007, 2008 Z RESEARCH Inc. <http://www.zresearch.com>
GlusterFS comes with ABSOLUTELY NO WARRANTY.
You may redistribute copies of GlusterFS under the terms of the GNU General
Public License.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20081211/6c3046fe/attachment.html>