Norman Gaywood | 1 Mar 2003 03:09
Picon
Picon
Favicon

Re: Linux Win, slow no matter what. Help would be appreciated....

On Fri, Feb 28, 2003 at 11:54:32PM +0100, Robert Lindgren wrote:
> During copy of one 500Mb file from /home/user to /home/user/test/
[snip]
> as you can see "w" took 2.55 secs to execute....

and you have 6Gig of memory. I bet this is the VM bug in 2.4.18. See:

https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=79257

--

-- 
Norman Gaywood -- School of Mathematical and Computer Sciences
University of New England, Armidale, NSW 2351, Australia
norm <at> turing.une.edu.au     http://turing.une.edu.au/~norm
Phone: +61 2 6773 2412     Fax: +61 2 6773 3312

_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge <at> dell.com
http://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq or search the list archives at http://lists.us.dell.com/htdig/

jason andrade | 1 Mar 2003 03:42

Re: Linux Win, slow no matter what. Help would be appreciated....

On Fri, 28 Feb 2003, Robert Lindgren wrote:

one other thing to try - can you turn off hyperthreading when you boot
the system ?

> Just to state for the record, this server is a freshly installed rh 8 with stock kernel, hw raid5, just me
accessing it and no other processes running.

ok, some other things to try - can you upgrade the kernel to the
latest errata version ?

most strange to be seeing these waits.  i assume you have upgraded
the raid firmware to whatever is latest of course.

regards,

-jason

_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge <at> dell.com
http://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq or search the list archives at http://lists.us.dell.com/htdig/

Norman Gaywood | 1 Mar 2003 04:09
Picon
Picon
Favicon

Re: Linux Win, slow no matter what. Help would be appreciated....

On Sat, Mar 01, 2003 at 01:09:16PM +1100, Norman Gaywood wrote:
> On Fri, Feb 28, 2003 at 11:54:32PM +0100, Robert Lindgren wrote:
> > During copy of one 500Mb file from /home/user to /home/user/test/
> [snip]
> > as you can see "w" took 2.55 secs to execute....
> 
> and you have 6Gig of memory. I bet this is the VM bug in 2.4.18. See:
> 
> https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=79257

Actually, after having a closer look at the information you sent, it's
probably something else. I'll take back my suggestion and withdraw to
be amongst the lurkers again :-)

--

-- 
Norman Gaywood -- School of Mathematical and Computer Sciences
University of New England, Armidale, NSW 2351, Australia
norm <at> turing.une.edu.au     http://turing.une.edu.au/~norm
Phone: +61 2 6773 2412     Fax: +61 2 6773 3312

_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge <at> dell.com
http://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq or search the list archives at http://lists.us.dell.com/htdig/

Titzer, David | 1 Mar 2003 04:15
Favicon

megaraid under SuSE SLES7


We've been trying to test an Oracle import on a PE8450 with a PV221 hanging
off a PERC3/DC. Things had been going fairly well (just not the imports, but
that's another story), until the test imports started taking forever.
Message log file showed hour-upon-hour of scsi timeouts on channel 2 of the
PERC3/DC. Nothing on channel 1. 

For the record, this is an Oracle 8.1.7.4 installation, running on SuSE
SLES7 (patched to be current), and a PowerEdge 8450 6-way Xeon with 8GB RAM.
Oracle calls this a certified platform. The timeouts--and resulting
slowdowns--are a sudden thing, because the system had been running nicely. 

We've been tinkering with this system for only a few weeks. Some updates
were applied early on, then I put the recent ones on today. 

Is this timeout issue a known problem, and is there a cure? 

As a separate question, is it possible to install dellmgr on SuSE SLES7? is
there a version that will install at all?

-dat

--
David A. Titzer
Systems Administrator
ServiceBench
410-290-3674
dtitzer <at> servicebench.com

_______________________________________________
(Continue reading)

Eric Phillips | 1 Mar 2003 05:11
Picon

compiling kernel

I have a dell poweredge 1600sc and put redhat 8 on it and was trying to compile a 2.4.20 kernel for it but keep getting a kernel panic at boot.  I think I'm missing the scsi controller which is the lsi ultra320 scsi controller or something is wrong in grub.  I also notice that I don't have a modules info file that corresponds to the new kernel.  I do have the .img and vmlinuz though.
 
Is there a good way to find out what drivers I need and where to get them so I can compile the kernel with the right drivers?
 
On an unrelated note, anyone here use the atalk services and use QuarkXpress?  I noticed there was an issue with that program saving files properly on the server and was wondering if someone knew a workaround?
 
Thanks everyone for your time
 
Eric :)
John Logsdon | 1 Mar 2003 10:49

More kernel queries

Hi

This is really directed to kernel hacking types, RedHat and Dell lurkers
on the list!

I want to compile in the latest patches of grsecurity that have a process
accounting system included but may I ask for advice first.  If anyone has
already done this, some specific guidance would be appreciated.

The system is PE2650/6GB (4+2 redundant to avoid the bigmen problems at
the moment) and RH8.0.

The grsecurity patches require 2.4.20 but the latest RH version is
2.4.18-24.7 at least on the errata page (3 Feb 03).

It is possible that some of the patches from 2.4.18 to 2.4.18-24.7 have
found their way into 2.4.20.

I would like if possible to retain any changes RH have put in and that
will have been tested by Dell.

The grsecurity patch will not work with 2.4.18-24.7 (I expect although I
haven't tried this!)

What is a suggested path?  

TIA

John

John Logsdon                               "Try to make things as simple
Quantex Research Ltd, Manchester UK         as possible but not simpler"
j.logsdon <at> quantex-research.com              a.einstein <at> relativity.org
+44(0)161 445 4951/G:+44(0)7768982349       www.quantex-research.com

_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge <at> dell.com
http://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq or search the list archives at http://lists.us.dell.com/htdig/

Marc Schmitt | 1 Mar 2003 12:10
Picon

Re: More kernel queries

Hi John,

John Logsdon wrote:

>Hi
>
>This is really directed to kernel hacking types, RedHat and Dell lurkers
>on the list!
>
>I want to compile in the latest patches of grsecurity that have a process
>accounting system included but may I ask for advice first.  If anyone has
>already done this, some specific guidance would be appreciated.
>
>The system is PE2650/6GB (4+2 redundant to avoid the bigmen problems at
>the moment) and RH8.0.
>
>The grsecurity patches require 2.4.20 but the latest RH version is
>2.4.18-24.7 at least on the errata page (3 Feb 03).
>
>It is possible that some of the patches from 2.4.18 to 2.4.18-24.7 have
>found their way into 2.4.20.
>
Look at the 2.4.18-24.7 SRPM, quite a bunch of patches come from the 
2.4.19 and up kernels and AC stuff.

>
>I would like if possible to retain any changes RH have put in and that
>will have been tested by Dell.
>
Sounds like a challenge, considering the amount of patches RH puts in... ;)

>
>The grsecurity patch will not work with 2.4.18-24.7 (I expect although I
>haven't tried this!)
>
>What is a suggested path?
>
Chances are very high that it will fail. I'm using EVMS and did not even 
bother trying to patch a RH kernel either. For what I needed (XFS and 
EVMS), I took the latest stable patchset (forces me to use kernel 
2.4.19), added the latest stable qla2xxx drivers, used the patches in 
2.4.18-24.7 to add e100 and e1000 support and probably will add the 
latest tg3 driver the next time I fiddle with it, as the latest results 
look very promising. All this for a PE4600.

I don't know what hardware the PE2650 has, therefore I can't tell you 
what 2.4.20 will be missing. Give it a try, patch 2.4.20, boot it and 
check it out.

Regards,
    Marc

_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge <at> dell.com
http://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq or search the list archives at http://lists.us.dell.com/htdig/

Marc Schmitt | 1 Mar 2003 12:24
Picon

Re: compiling kernel

Hi Eric,

Have a look at /etc/modules.conf, it contains the list of drivers 
(modules) that the system needs to boot from a SCSI controller.
Use mkinitrd to create the initrd image ('mkinitrd 
/boot/initrd-2.4.20.img 2.4.20') and add this to grub. If kernel 2.4.20 
has a driver for you SCSI controller, mkinitrd will put it into 
initrd-2.4.20.img, now the kernel will be able to load it *before* it 
tries to switch to the root FS on the SCSI disk => no more kernel panic.

HTH

Greetz
    Marc

Eric Phillips wrote:

> I have a dell poweredge 1600sc and put redhat 8 on it and was trying 
> to compile a 2.4.20 kernel for it but keep getting a kernel panic at 
> boot.  I think I'm missing the scsi controller which is the lsi 
> ultra320 scsi controller or something is wrong in grub.  I also notice 
> that I don't have a modules info file that corresponds to the new 
> kernel.  I do have the .img and vmlinuz though.
>  
> Is there a good way to find out what drivers I need and where to get 
> them so I can compile the kernel with the right drivers?
>  
> On an unrelated note, anyone here use the atalk services and use 
> QuarkXpress?  I noticed there was an issue with that program saving 
> files properly on the server and was wondering if someone knew a 
> workaround?
>  
> Thanks everyone for your time
>  
> Eric :)

_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge <at> dell.com
http://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq or search the list archives at http://lists.us.dell.com/htdig/

Seth Mos | 1 Mar 2003 14:38
Picon
Picon
Favicon

Re: Linux Win, slow no matter what. Help would be appreciated....

On Fri, 28 Feb 2003, Robert Lindgren wrote:

> On Sat, 1 Mar 2003 07:37:48 +1000 (EST)
> jason andrade <jason <at> rtfmconsult.com> wrote:
>
> > On Fri, 28 Feb 2003, Robert Lindgren wrote:
> >
> > > The server I run the tests on was a 2650, dual 2.4, 6gb ram, 3 fujitso map3735nc 73GB drives, and the
perc3/Di/adaptec 7889/adaptec 29160 cards.
> >
> > as seth has suggested, turn off the 6G and perhaps configure the additional 2G as
> > "sparebank"
>
> I did actually run most of the tests with just 1GB of ram to be able to save some time.

Ah, but booting with mem=1024 instead of recompiling the kernel without
PAE support will still result in skewed results. File servers and PAE
don't seem to mix that well. Use PAE and the full 6GB when you are running
databases and you need the extra memory.

> >  also, make sure you are running on the latest version of the
raid firmware.
>
> Yes it's the latest version, just got this server a couple of days ago.

Ahem, getting your server a few days ago does not mean it actually has the
latest firmware. Better check this if you did not already do so. It should
be something like 2.7.1 for adaptec based raid controllers.

> As I said the where not that much difference regardless what config I ran, when it comes to write
performance. Regarding read ext3 (data=journal) was the slowest, reiserfs little faster and ext3
(data=writeback) fastest. Can post some figures later when I've got more time if it is of any interest.

Have you tried using XFS for the data partition? I use XFS on production
machines for over a year without significant problems.

The performance should be alright. A few days ago a post to the list was
made with the performance differences for XFS vs ext3 in a renderfarm
where it was used as the NFS fileserver. You can check the archives.

If you want to try it, you can find the new XFS 1.2 release from the
website http://oss.sgi.com/projects/xfs/ and if you need errata kernels
with the XFS 1.2 patch you can find those on my site
http://iserv.nl/files/xfs/

Don't forget fetching the userspace tools from the oss website.

> Yes the system is unresponsive with hw raid, when in I/O bottleneck has kicked in. Can run any program like
ps,w (or anything else for that matter), sure I'm to run it after a couple of sec (5-15sec), same in windows
pressed the start menu, took 10-15 sec for it to pop up.

My sixth sense says that it has to do with the Highmem support. Have you
tried actually taking the extra ram and sticking to 4GB?

Cheers
Seth

_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge <at> dell.com
http://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq or search the list archives at http://lists.us.dell.com/htdig/

Tom Rawson | 1 Mar 2003 15:21

Slackware 8.1 on Poweredge 4100/200 w/ PERC

Just posted this to the web-based forums but now I see this list, which 
looks like a better venue ...

I've got one of these systems and I'm trying to install Linux 
(Slackware 8.1 / kernel 2.4.18). The RAID controller is the AMI 
(megaraid) PERC, 4 each 9 GB disks configured as 27 GB RAID 5 array, 
all as a single drive which I will partition within Linux.  

I have read a lot of online sources and am quite confused about how to 
find the right drivers and boot disk image to get it up and running -- 
partly because the documents all refer to Red Hat version numbers.

The standard adaptec.s boot disk appears to recognize the onboard 7860 
and PCI 7880 controllers, and the SCSI CD-ROM, but I can't see the RAID 
drive in /dev once booted. I'm not sure if the problem is that I need 
to update support for the RAID controller (though the reading I've done 
suggests it is supported in 2.4.18) or if there is a configuration 
problem elsewhere.

I see the megaraid.s boot disk available from slack, but its 
description is " Supports AMI MegaRAID 428 and 438 (and maybe 466), 
SCSI host adapters." which does not sound like the 4100 PERC -- plus I 
need to boot with support for BOTH the Adaptec and PERC controllers -- 
right?

Thanks for any help ...

 ----------
 Tom Rawson

_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge <at> dell.com
http://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq or search the list archives at http://lists.us.dell.com/htdig/


Gmane