http://www.linuxtoday.com/high_performance/2007022803926OSDPHE
"Today's new breed of smaller, cheaper mainframes, paired with the Linux
operating system, look like an attractive alternative to Unix on RISC or
SPARC servers. Linux on the mainframe seems to give us the best of all
worlds: the depend
Use google to search on "Linux Symposium" from last year. It had a paper
written by guys at SGI that compared various filesystems. It is Intel
based but the lessons are applicable to zSeries. You most definitely do
not want reiser, especially in an SMP environment: reiser will try and
get hold of t
Was wondering if anyone here has done some performance testing with
Mysql databases and various filesystem types, like ext3, reiserfs, xfs
and so on. I've seen some tendencies on google searches toward reiserfs
for Mysql, but that's on Intel physical systems. Given that reiserfs
has more cpu over
>>> On Wed, Feb 28, 2007 at 4:25 PM, in message <[EMAIL PROTECTED]>,
Bennie Hicks <[EMAIL PROTECTED]> wrote:
> I'm not a SLES or YaST user, but in RHEL I have come behind folks that
> set the hostname/domain in kernel parms (/etc/sysctl.conf). Probably not
> related, just mentioning what I have s
>>> On Wed, Feb 28, 2007 at 5:09 PM, in message
<[EMAIL PROTECTED]>,
James Melin <[EMAIL PROTECTED]> wrote:
> Hmm. I'm already pointed to you/novell.com on that machine.
>
> Update Configuration─┐
> Installation source │
> User- Defined Loca>┴
Hmm. I'm already pointed to you/novell.com on that machine.
Update Configuration─┐
Installation source │
User-Defined Loca>┴ │
Location │
https://you.novell.com/update/▒▒ │
[New Serve
Murkier? Great. I'll try that against the you.novell.com and see what happens.
Mark Post <[EMAIL PROTECTED]>
Sent by: Linux on 390 Port
To
>>> On Wed, Feb 28, 2007 at 2:17 PM, in message
<[EMAIL PROTECTED]>, Mark Pace
<[EMAIL PROTECTED]> wrote:
> On 2/28/07, Romanowski, John (OFT) <[EMAIL PROTECTED]> wrote:
>>
>> Can someone share how they mirror the SLES9 patches locally?
>>
>> I'd be interested in this also.
It's done exactly the
For the purposes talked about in this case, the "free" command is
sufficient.
Mark Post
-Original Message-
From: Linux on 390 Port [mailto:[EMAIL PROTECTED] Behalf Of
barton
Sent: Wednesday, February 28, 2007 2:08 PM
To: LINUX-390@VM.MARIST.EDU
Subject: Re: Vdisk
I think there is a bet
I'm not a SLES or YaST user, but in RHEL I have come behind folks that
set the hostname/domain in kernel parms (/etc/sysctl.conf). Probably not
related, just mentioning what I have seen before that got me on some
manual clones.
Mark Post wrote:
On Wed, Feb 28, 2007 at 12:14 PM, in message
<[
On 2/28/07, Romanowski, John (OFT) <[EMAIL PROTECTED]> wrote:
Can someone share how they mirror the SLES9 patches locally?
I'd be interested in this also.
--
Mark Pace
Mainline Information Systems
--
For LINUX-390 subscribe
I think there is a better opportunity for a performance monitor that
doesn't cost "obscene" amounts of resource, and reports correctly
(Oh yah, ESALPS comes to mind)
Mark Post wrote:
On Wed, Feb 28, 2007 at 11:29 AM, in message
<[EMAIL PROTECTED]>, Adam Thornton
<[EMAIL PROTECTED]> wrote
Can someone share how they mirror the SLES9 patches locally?
I'm interested in which command (rsync, wget, ?) and which parameters,
I'd run to mirror the SLES9 internet-based patch site onto local disk. I
have a Novell SLES9 maintenance contract and access to the site.
Before Novell bought Suse
On 2/28/07, James Melin <[EMAIL PROTECTED]> wrote:
I'm pointing at novell currently, and I ran that update command. Still no
change on the SPident results.
James -
Try the SPindent -
I had the same issue. Once I ran this I found a software product not being
updated automatically. I manu
I am presuming you are pointing at the novell install servers or a local copy
you maintain by daily update?
I'm pointing at novell currently, and I ran that update command. Still no
change on the SPident results. Obviously I am missing something on how to
make YOU find the SP3 patches. If you ha
Thanks, Mark. I see my error was in assuming lower case on the hostname
file. I found it thanks to your suggestion. I'm still not sure why YaST
hangs, but I am at least able to change the name manually.
Regards,
Dave
-Original Message-
From: Linux on 390 Port [mailto:[EMAIL PROTECTED] On
>>> On Wed, Feb 28, 2007 at 12:14 PM, in message
<[EMAIL PROTECTED]>, Dave Keeton
<[EMAIL PROTECTED]> wrote:
-snip-
> Has anyone else seen this? Does anyone know where SuSE stores its
> hostname variable (other than /etc/hostname which doesn't appear to be
> read by YaST). I can use the "hostname
I'm seeing an issue when I try to change the hostname on cloned SLES10
guests. When making changes to the network config through YaST, I am
able to change the name, but when YaST goes to save the changes, it
hangs on "Write hostname and DNS configuration". It never completes.
Has anyone else seen
>>> On Wed, Feb 28, 2007 at 11:29 AM, in message
<[EMAIL PROTECTED]>, Adam Thornton
<[EMAIL PROTECTED]> wrote:
> On Feb 28, 2007, at 9:53 AM, Levy, Alan wrote:
>
>> When I do a TOP, the Performance Toolkit shows a spike to 90% of the
>> cpu. When I ctrl- C out of top, it goes down to 32% (normal)
On Feb 28, 2007, at 9:53 AM, Levy, Alan wrote:
When I do a TOP, the Performance Toolkit shows a spike to 90% of the
cpu. When I ctrl-C out of top, it goes down to 32% (normal).
This one server is a memory hog. It uses Tomcat (on sles9 sp3) with
350
connections at one time (I do not own this ser
When I do a TOP, the Performance Toolkit shows a spike to 90% of the
cpu. When I ctrl-C out of top, it goes down to 32% (normal).
This one server is a memory hog. It uses Tomcat (on sles9 sp3) with 350
connections at one time (I do not own this server, I just set it up and
we are currently rearchi
On Feb 28, 2007, at 9:28 AM, Levy, Alan wrote:
Unfortunately, I cannot do a TOP. It blows my server away.
In the words of the inestimable Mr. Loaf: "Stop right there!"
This is indicative of something very, very seriously wrong. I mean,
*very* wrong.
Now, maybe all you meant is the relativel
>>> On Wed, Feb 28, 2007 at 10:17 AM, in message
<[EMAIL PROTECTED]>,
James Melin <[EMAIL PROTECTED]> wrote:
> Perhaps I am mistaken, but is it not possible to use Yast Online Update to
> apply SP3?
I always do it from the command line:
online_update -l en_us security recommended patchlevel docu
One good thing I like on SUSE distributions is the documentation on
directory /docu on first instalation CD. I dont't saw it on others distros.
Until Sles9 many Redbooks was included. Very usefull.
But on Sles10 (at lest on a copy I saw), only few Suse manuals are
available.
__
On 2/28/07, James Melin <[EMAIL PROTECTED]> wrote:
I put on all recommended maintenance to a test system but now SPident is
saying :
CONCLUSION: System is NOT up-to-date!
foundSLES-9-s390x-SP2 + "online updates"
expected SLES-9-s390x-SP3
try SPident - <- 4 v's This will b
Unfortunately, I cannot do a TOP. It blows my server away. Here's a
free:
total used free sharedbuffers
cached
Mem: 15509841547956 3028 0 1440
34324
-/+ buffers/cache:1512192 38792
Swap: 1583792 2207961362996
-
Of the 30 servers I have now, only 3 are currently swapping (and not
much).
Btw: I got the 2 x memory algorithm from IBM and I've been using it for
about 4 years now. I am not having any problems with 1G swap on dasd but
lately have seen many people talking about vdisk as a replacement.
I can ge
> > Many of the 30 servers run intense java applications. Some of them
are
> > at 2G logon memory and they still swap.
> > What would you recommend as swap space for these users ?
Can you post what a sample server is actually using? Example (from 'top'
on a lightly loaded guest):
Mem: 1036016k
A mod 9 is ~7GB, 3 mod3's is the same as a mod 9. So you have ~21GB.
Mark Post wrote:
On Wed, Feb 28, 2007 at 9:55 AM, in message
<[EMAIL PROTECTED]>, "Levy,
Alan" <[EMAIL PROTECTED]> wrote:
Thanks for the answer.
1) Make sure you have sufficient paging space to back CP allocating 1G
of pag
And from practical experience I say be generous with paging space. The day you
will be glad having it will probably come sooner than later.
Best regards,
Pieter Harder
[EMAIL PROTECTED]
tel +31-73-6837133 / +31-6-47272537
>>> [EMAIL PROTECTED] 28-02-2007 16:15 >>>
>>> On Wed, Feb 28, 2007 at
> I have one mod9 and 6 mod3 for paging (total pages 5281K) - abo0ut 38%
> used right now. Would this be sufficient ?
Hmm. Without knowing the usage patterns of your system, and not having
perf data, I'd be cautious in saying it's enough for total safety, but
it looks like enough to get started w/
Perhaps I am mistaken, but is it not possible to use Yast Online Update to
apply SP3?
I put on all recommended maintenance to a test system but now SPident is saying
:
CONCLUSION: System is NOT up-to-date!
foundSLES-9-s390x-SP2 + "online updates"
expected SLES-9-s390x-SP3
the sles-rele
Please refer to:
http://www.ibm.com/developerworks/linux/linux390/whatsnew.html
for the 2007-02-28 change summary:
1) "October 2005 stream":
> Recommended kernel 2.6.16 patches:
- Patch 13 with new functionality
- optional add-on patch Kernel NSS Support
- optional add-on patch S
>>> On Wed, Feb 28, 2007 at 9:55 AM, in message
<[EMAIL PROTECTED]>, "Levy,
Alan" <[EMAIL PROTECTED]> wrote:
> Thanks for the answer.
>
> 1) Make sure you have sufficient paging space to back CP allocating 1G
> of pages per guest. The initial allocation that comes with VM isn't big
> enough, and
>>> On Wed, Feb 28, 2007 at 9:57 AM, in message
<[EMAIL PROTECTED]>, "Levy,
Alan" <[EMAIL PROTECTED]> wrote:
> Many of the 30 servers run intense java applications. Some of them are
> at 2G logon memory and they still swap.
>
> What would you recommend as swap space for these users ?
You would
Look at your disk swap utilization (not allocation) for those servers
and use that much for vdisk. Of course you may want to add some for any
storage need spikes.
Levy, Alan wrote:
Many of the 30 servers run intense java applications. Some of them are
at 2G logon memory and they still swap.
Wh
>>> On Wed, Feb 28, 2007 at 7:45 AM, in message
<[EMAIL PROTECTED]>, "Lee, Gary D."
<[EMAIL PROTECTED]> wrote:
> Sir:
>
> That seemed to work.
> The module is now loaded.
> Now I think I need to create this mapping information.
> I've got everything I need except the lun created by the san.
> I
Many of the 30 servers run intense java applications. Some of them are
at 2G logon memory and they still swap.
What would you recommend as swap space for these users ?
Alan Levy
VM/Linux Administrator
W: 718-403-8020
C: 347-401-4629
-Original Message-
From: Linux on 390 Port [mailto:[E
Thanks for the answer.
1) Make sure you have sufficient paging space to back CP allocating 1G
of pages per guest. The initial allocation that comes with VM isn't big
enough, and while you won't use it all at once, you WILL eventually need
it.
I have one mod9 and 6 mod3 for paging (total pages 528
The lpar that i am using is the same one i used for the sles9/zvm. I
actually shutdown the zvm lpar and then do a LOAD from CDrom for this same
lpar.
So it is strange that i still get the "dasd is not active". I checked the
other lpars and made sure that all other lpars have the dasd addresses (
>>> On Wed, Feb 28, 2007 at 9:27 AM, in message
<[EMAIL PROTECTED]>, Caleb C
Ong <[EMAIL PROTECTED]> wrote:
> Just to clarify, I am ipling from the HMC cdrom (using sles9 cd1) . after
> ipl i setup the osa network and using samba to window server.
> After i start ssh session, i logon as root to
On 2/28/07, Caleb C Ong <[EMAIL PROTECTED]> wrote:
Just to clarify, I am ipling from the HMC cdrom (using sles9 cd1) . after
ipl i setup the osa network and using samba to window server.
After i start ssh session, i logon as root to start YAST. sorry , wasn't
so clear about this.
So the confi
Just to clarify, I am ipling from the HMC cdrom (using sles9 cd1) . after
ipl i setup the osa network and using samba to window server.
After i start ssh session, i logon as root to start YAST. sorry , wasn't
so clear about this.
Caleb C. Ong
IBM Global Services
IT-Specialist
Tel :995-2131
On 2/28/07, Caleb C Ong <[EMAIL PROTECTED]> wrote:
I tried to manually online the device using :
echo 1 > sys/bus/ccw/devices/0.0.4549/online
when i do a cat /sys/bus/ccw/devices/0.0.4549/online
it always shows 0.
You may also want to verify that the disks are in fact connected to your
virt
Strictly speaking, yes. But, if you don't use it all, you can get away with
overcommitting to some degree. Your performance monitor can tell you how much
you can get away with.
-Original Message-
From: "Harold Grovesteen" <[EMAIL PROTECTED]>
To: "LINUX-390@VM.MARIST.EDU"
Sent: 2/28/07
Caleb C Ong <[EMAIL PROTECTED]> wrote on 28.02.2007 12:07:43:
> when i do a cat /sys/bus/ccw/devices/0.0.4549/online
>
> it always shows 0.
Make sure the dasd modules are loaded (dasd_mod, dasd_diag_mod,
dasd_fba_mod, dasd_eckd_mod) and try again. If that doesn't help look in
the output of 'dme
On 2/28/07, Harold Grovesteen <[EMAIL PROTECTED]> wrote:
The swap space will not be reclaimed, just moved from the Linux guests
to VM, correct?
Yes, you need to add paging disks to z/VM to support the increased
demand through the VDISK. For good paging performance you need to
remain under 50%
I am trying to install sles9 on a 3390 disk. I've already installed sles9
on top of vm and this is working already.
I logon as root, start YAST.
When i go thru yast panels, i tried to "ACTIVATE" the two 3390 disks that i
will be using. There is no error message. when i tried to format the dasd,
From Alan Levy:
It would be nice to reclaim the 1G of swap dasd space for the user to use.
David Boyes wrote:
You definitely need to be sure you have CP paging
space sufficient to back all the allocated VDISK you define, though --
that's just a good safety net.
The swap space will not be r
Rob van der Heij wrote:
I know someone who runs multiple SAP systems in production with 16 GB
of swap space each, primarily to satisfy the vendor requirements.
Because of CP limitations they come in portions of 2 GB, so only the
first one is being used to swap, if any. He monitors the usage of th
On 2/28/07, Mark Perry <[EMAIL PROTECTED]> wrote:
I will be experimenting over the next few months (with SLES10) to see
what "really" works for SAP on zLinux under z/VM using VDISK.
I know someone who runs multiple SAP systems in production with 16 GB
of swap space each, primarily to satis
Mark Post wrote:
Why do you have 1GB of swap defined for each guest? That sounds far too high for most uses. Are
you following the rule of thumb to use "twice the amount of physical RAM?" If so, you
shouldn't be doing that. It's not even valid in the midrange world any more, although a lot
52 matches
Mail list logo