Dear Chris,
I just read the ChangeLog for 2.4.19-pre4 and saw, that there were
several problems concerning the interaction of NFS with VFS addressed by
recent patches. Furthermore, the 1GBit-driver ns83820 which we use has
been patched. So hopefully my problems will be solved in the near fu
Thanks Chris for your understanding, we are really open here for new
developments (otherwise we would not use a linux server for
production...).
If you have approved your patches, I will surely give them a try since up
to know I did not experience any problems with my self-compiled 2.4.18
ker
On Monday, March 18, 2002 02:54:51 PM +0100 [EMAIL PROTECTED] wrote:
> Dear Chris,
>
> Thanks for your reply, as I mentioned I don't have a second GBit-server
> here and our original machine is a production environment, so I can't
> undertake all the test.
Not a problem, I was hoping you
Dear Chris,
Thanks for your reply, as I mentioned I don't have a second GBit-server
here and our original machine is a production environment, so I can't
undertake all the test.
I don't even may reproduce my first test during the day, bcause the people
here might lose their samba-hosted file
On Monday, March 18, 2002 10:20:28 AM +0300 Oleg Drokin <[EMAIL PROTECTED]> wrote:
> So this one is slightly better.
> And after all I tried some internal (soon to be released) speedup patches
> from Chris and got 3m38.718s time which is 9.17 megabytes per second.
>
I think these speedup pa
On Mon, 18 Mar 2002 06:16:39,Valdis Kletnieks wrote :
> On Mon, 18 Mar 2002 09:11:14 +0300, Oleg Drokin said:
>
> > It is somehow related to knfsd, I think. I will tell you what I will be
> > able to find.
>
> > The only thing that ext{2,3}/reiserfs differs right now is buffer flushing
> > policy,
Dear Nikita,
No, we don't use sync nfs mode, but we transfer files, which are larger
than the physical memory available and we have reduced the lower limit for
the perecentage of dirty buffer for activating bdflush in
/proc/sys/vm/bdflush from 30% to 5%, because the hardware RAID system is
[EMAIL PROTECTED] writes:
> Dear Oleg,
>
> The transfer rate is limited by the clients, which are connected to a
> 100MBit network over the 1GBit-->8x 100MBit switch and by the hardware
> raid controller we use.
> The performance is 8MB/s and stable, if we use ext3 and 6MB/s and very
>
Dear Oleg,
The transfer rate is limited by the clients, which are connected to a
100MBit network over the 1GBit-->8x 100MBit switch and by the hardware
raid controller we use.
The performance is 8MB/s and stable, if we use ext3 and 6MB/s and very
unstable (as outlined earlier), if we use reise
Hello!
On Mon, Mar 18, 2002 at 10:29:32AM +0100, [EMAIL PROTECTED] wrote:
> OK, I have another box with a usual setup: eepro100 network card (100
> MBit/s) and I can compare kernel 2.4.16 and kernel 2.4.18 on this machine.
> But actually it sems to me, that there are some effects, that arise
Dear Oleg,
OK, I have another box with a usual setup: eepro100 network card (100
MBit/s) and I can compare kernel 2.4.16 and kernel 2.4.18 on this machine.
But actually it sems to me, that there are some effects, that arise only
with the 1GBit network and I don't have a second GBit server at
Hello!
On Mon, Mar 18, 2002 at 09:39:11AM +0100, [EMAIL PROTECTED] wrote:
> Unfortunately, I cant't easily test undertake various test series, because
> this is a production system and we have mostly converted our reiser
> partitions to ext3 for stability reasons. Now, that you encourage me, I
Hi Oleg,
I don't have any network card errors in the output of ifconfig. The only
suspicious thing is the error message from samba, which is included in my
original posting.
Unfortunately, I cant't easily test undertake various test series, because
this is a production system and we have mo
Hello!
On Mon, Mar 18, 2002 at 08:53:56AM +0100, [EMAIL PROTECTED] wrote:
> Thanks for your additional investigations, since I use a 2.4.18 client
> and a 2.4.16 server and do not use any special mount options for NFS, I
> use NFSv3. Our peek performance is not the real problem here, I think
Dear Oleg,
Thanks for your additional investigations, since I use a 2.4.18 client
and a 2.4.16 server and do not use any special mount options for NFS, I
use NFSv3. Our peek performance is not the real problem here, I think the
blocks on the network interface are the things, that really hurt.
Hello!
On Mon, Mar 18, 2002 at 09:11:14AM +0300, Oleg Drokin wrote:
> > When the network blocks as I outlined above, I'm unable to enter a command
> > on the ssh prompt from the server for around 1 second and output of top et
> > al. is delayed.
> During my recent checks I found that using
16 matches
Mail list logo