Hi, I just installed the last version of glusterfs 3.0 and I got really bad performance. Here it is : # dd if=/dev/zero of=/users/glusterfs_mnt/sample bs=1k count=100000 This create a file of 100Mo and I got those results : NFS = 75 Mo/s Gluster=5,8Mo/s I tried to change block size, change value of write behind parameters, to add Read Ahead translator, to remove all performance translator( and it's worst :\ ), try with afr translator ... but no change! My configuration is : replicated and distributed (RAID 10 over network) on server side over 4 bricks, Giga Ethernet on all servers and clients What kind of performance do you have? Is it normal? I also tried to run a iostat test and it run during 24H of CPU time (3 days at all).... Here my vol file of server and client computer : # /etc/glusterfs/glusterfsd.vol volume posix type storage/posix option directory /users/gluster-data end-volume volume locks type features/locks subvolumes posix end-volume volume brick type performance/io-threads option thread-count 8 subvolumes locks end-volume volume server type protocol/server option transport-type tcp/server option auth.addr.brick.allow * subvolumes brick end-volume ------------------------------------------------ #/etc/glusterfs/glusterfs.vol volume brick1 type protocol/client option transport-type tcp option remote-host # IP address of the remote brick option remote-subvolume brick # name of the remote volume end-volume volume brick2 type protocol/client option transport-type tcp option remote-host # IP address of the remote brick option remote-subvolume brick # name of the remote volume end-volume volume brick3 type protocol/client option transport-type tcp option remote-host option remote-subvolume brick end-volume volume brick4 type protocol/client option transport-type tcp option remote-host option remote-subvolume brick end-volume volume rep1 type cluster/replicate subvolumes brick1 brick2 end-volume volume rep2 type cluster/replicate subvolumes brick3 brick4 end-volume volume distribute type cluster/distribute subvolumes rep1 rep2 end-volume volume writebehind type performance/write-behind option window-size 1MB subvolumes distribute end-volume volume cache type performance/io-cache option cache-size 512MB subvolumes writebehind end-volume _________________________________________________________________ T?l?chargez Internet Explorer 8 et surfez sans laisser de trace ! http://clk.atdmt.com/FRM/go/182932252/direct/01/
You can try this: # /etc/glusterfs/glusterfsd.vol volume server type protocol/server option transport-type tcp/server * option transport.socket.nodelay on* option auth.addr.brick.allow * subvolumes brick end-volume and on each brick: #/etc/glusterfs/glusterfs.vol volume brick1 type protocol/client option transport-type tcp *option transport.socket.nodelay on* option remote-host # IP address of the remote brick option remote-subvolume brick # name of the remote volume end-volume Good luck Andoni Ayala 2009/12/21 anthony garnier <sokar6012 at hotmail.com>> > Hi, > I just installed the last version of glusterfs 3.0 and I got really bad > performance. > Here it is : > # dd if=/dev/zero of=/users/glusterfs_mnt/sample bs=1k count=100000 > > This create a file of 100Mo and I got those results : > NFS = 75 Mo/s > Gluster=5,8Mo/s > > I tried to change block size, change value of write behind parameters, to > add Read Ahead translator, to remove all performance translator( and it's > worst :\ ), try with afr translator ... but no change! > > My configuration is : replicated and distributed (RAID 10 over network) on > server side over 4 bricks, Giga Ethernet on all servers and clients > > What kind of performance do you have? Is it normal? > I also tried to run a iostat test and it run during 24H of CPU time (3 days > at all).... > Here my vol file of server and client computer : > > # /etc/glusterfs/glusterfsd.vol > > volume posix > type storage/posix > option directory /users/gluster-data > end-volume > > volume locks > type features/locks > subvolumes posix > end-volume > > volume brick > type performance/io-threads > option thread-count 8 > subvolumes locks > end-volume > > volume server > type protocol/server > option transport-type tcp/server > option auth.addr.brick.allow * > subvolumes brick > end-volume > ------------------------------------------------ > #/etc/glusterfs/glusterfs.vol > > volume brick1 > type protocol/client > option transport-type tcp > option remote-host # IP address of the remote brick > option remote-subvolume brick # name of the remote volume > end-volume > > volume brick2 > type protocol/client > option transport-type tcp > option remote-host # IP address of the remote brick > option remote-subvolume brick # name of the remote volume > end-volume > > volume brick3 > type protocol/client > option transport-type tcp > option remote-host > option remote-subvolume brick > end-volume > > volume brick4 > type protocol/client > option transport-type tcp > option remote-host > option remote-subvolume brick > end-volume > > volume rep1 > type cluster/replicate > subvolumes brick1 brick2 > end-volume > > volume rep2 > type cluster/replicate > subvolumes brick3 brick4 > end-volume > > volume distribute > type cluster/distribute > subvolumes rep1 rep2 > end-volume > > volume writebehind > type performance/write-behind > option window-size 1MB > subvolumes distribute > end-volume > > volume cache > type performance/io-cache > option cache-size 512MB > subvolumes writebehind > end-volume > > > > > _________________________________________________________________ > T?l?chargez Internet Explorer 8 et surfez sans laisser de trace ! > http://clk.atdmt.com/FRM/go/182932252/direct/01/ > _______________________________________________ > Gluster-users mailing list > Gluster-users at gluster.org > http://gluster.org/cgi-bin/mailman/listinfo/gluster-users > >
I have already increased the blocksize of my dd=....... and I got no change.> Hi anthony, > > Increase the block size to get better performance > > > > Regards > Raghavendra Bhat_________________________________________________________________ Tchattez en direct en en vid?o avec vos amis ! http://www.windowslive.fr/messenger/
Yea of course, after each change I reload glusterfsd on server side and I remount partition on client side.> Date: Tue, 22 Dec 2009 09:18:38 +0900 > From: berenger at riken.jp > To: sokar6012 at hotmail.com > Subject: Re: [Gluster-users] performance issue > > anthony garnier wrote: > > I just tried to add your option but it change nothing :( > > Did you remount glusterfs partitions after this change and > before retrying? > On server and nodes? > > If this is the problem, we can send to the mailing list after. > > Regards, > F. >_________________________________________________________________ Windows 7 ? 35? pour les ?tudiants?! http://www.windows-7-pour-les-etudiants.com