[Ocfs2-users] Slow concurrent actions on the same LVM logical volume

[Arnold Maderthaner]

Hi !

I watched the servers today and noticed that the server load is very high
but not much is working on the server:
server1:
 16:37:11 up 32 min,  1 user,  load average: 6.45, 7.90, 7.45
server2:
 16:37:19 up 32 min,  4 users,  load average: 9.79, 9.76, 8.26
i also tried to do some selects on the oracle database today.
to do a "select count(*) from xyz" where xyz has 120 000 rows it took me
about 10 sec.
always the process who does the filesystem access is the first one.
and every command (ls,du,vgdisplay,lvdisplay,dd,cp,....) which needs access
to a ocfs2 filesystem seams like blocking or it seams to me that there is a
lock.
Can anyone please help me ?

yours

Arnold

On 8/22/07, Arnold Maderthaner <arnold.maderthaner at j4care.com> wrote:
>
> There is not really a high server load when writing to the disks but will
> try it asap. what do you mean with AST ?
> Can I enable some debuging on ocfs2 ? We have 2 Dell servers with about
> 4gb of ram each and 2 Dual Core CPUs with 2GHZ each.
> Can you please describe which tests todo again ?
>
> Thx for your help
>
> Arnold
>
> On 8/22/07, Sunil Mushran < Sunil.Mushran at oracle.com> wrote:
> >
> > Repeat the first test. This time run top on the first server. Which
> > process is eating the cpu? From the time it appears that the second
> > server is waiting for the AST that the first server is slow to send. And
> > it could be slow because it may be busy flushing the data to disk.
> > How much memory/cpu do these servers have?
> >
> > As far as the second test goes, what is the concurrent write performance
> > directly to the LV (minus fs).
> >
> > On Wed, Aug 22, 2007 at 01:14:02PM +0530, Arnold Maderthaner wrote:
> > > Hi 2 all !
> > >
> > > I have problems with concurrent filesystem actions on a ocfs2
> > > filesystem which is mounted by 2 nodes. OS=RH5ES and OCFS2=1.2.6
> > > F.e.: If I have a LV called testlv which is mounted on /mnt on both
> > > servers and I do a "dd if=/dev/zero of=/mnt/test.a bs=1024
> > > count=1000000" on server 1 and do at the same time a du -hs
> > > /mnt/test.a it takes about 5 seconds for du -hs to execute:
> > > 270M     test.a
> > >
> > > real    0m3.627s
> > > user    0m0.000s
> > > sys     0m0.002s
> > >
> > > or even longer. If i do a du -hs ond a file which is not written by
> > > another server its fast:
> > > 977M     test.b
> > >
> > > real    0m0.001s
> > > user    0m0.001s
> > > sys     0m0.000s
> > >
> > > If I write 2 different files from both servers to the same LV I get
> > > only 2.8mb/sec.
> > > If I write 2 different files from both servers to a different LV
> > > (testlv,test2lv) then I get about 30mb/sec on each write (so 60mb/sec
> > > combined which is very good).
> > > does anyone has an idea what could be wrong. Are there any mount
> > > options for ocfs2 which could help me? is this a lvm2 or ocfs2 problem
> >
> > > (locking,blocking,.... issue) ?
> > >
> > > Any help will be appriciated.
> > >
> > > yours
> > >
> > > Arnold
> > > --
> > > Arnold Maderthaner
> > > J4Care Inc.
> > >
> > > _______________________________________________
> > > Ocfs2-users mailing list
> > > Ocfs2-users at oss.oracle.com
> > > http://oss.oracle.com/mailman/listinfo/ocfs2-users
> >
>
>
>
> --
> Arnold Maderthaner
> J4Care Inc.




-- 
Arnold Maderthaner
J4Care Inc.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://oss.oracle.com/pipermail/ocfs2-users/attachments/20070823/c0530303/attachment.html