I'm about to get a new dual opteron machine and move to Suse 9.1
Professional from Debian. The Suse will come pre-installed. I'm new to
both smp and 64 bit, as well as Suse.
I wonder if you could point me to some documentation that tells me which
apps in the distribution are compiled as 64 bit - shall I need to
compile them all myself, or are they all 64 bit? Will any 32 bit apps
automatically be installed by a standard Suse install on that kind of
machine? Ideally, with a completely clean machine, I'd like to try to
keep the apps I use to 64 bit if it's possible (though I understand
there will be 32 bit libraries) or is this just a pipe dream?
Thanks.
--
Richard Kimber
http://www.psr.keele.ac.uk/
Hi all,
As promised, here is a summary of my experience so far with SuSE Professional
9.1 AMD64.
Best regards
---ooo---
Installation.
1. Default installation died with a segfault after installing software for
about 20 minutes.
Tried again with "apm=off acpi=off" and installation then succeeded through to
initial boot.
Used "apm=off acpi=off" on initial boot.
Connection to YOU succeeded.
During remainder of installation the system abruptly rebooted.
Starting YaST after reboot caused another abrupt reboot.
Rebooted using "noapic agp=off acpi=off apm=off".
Ran YaST in Curses mode, and selected YOU.
Updated software and updated the kernel to latest 2.6.4.
[Stability problems have so far disappeared with 2.6.5 kernel.
No longer need "noapic agp=off acpi=off apm=off".]
2. Display blinking and shuddering in framebuffer mode during bootup.
[This seems to have stopped after I updated the kernel to 2.6.5.]
3. Installing TrueType fonts apparently hangs.
[Fixed by editing fetchmsttfonts to prepend Japanese keihanna mirror to list
of mirrors.]
>diff -ub /var/lib/YaST2/you/scripts/fetchmsttfonts /root/bin/fetchmsttfonts
--- /var/lib/YaST2/you/scripts/fetchmsttfonts 2004-05-29 12:22:17.000000000
+1000
+++ /root/bin/fetchmsttfonts 2004-05-29 12:32:37.000000000 +1000
@@ -61,7 +61,7 @@
pushd $tmpdir &> /dev/null
for font in $FONTS; do
- for i in aleron belnet csnet easynews flow heanet switch twtelecom umn unc;
do
+ for i in keihanna aleron belnet csnet easynews flow heanet switch twtelecom
umn unc; do
archive=http://$i.$font
file=`echo $archive|awk -F "/" '{print $NF}'`
rm -f $file
---ooo---
Update to KDE 3.2.2 to match my previous SuSE 9.0 setup.
Update to KDE 3.2.2 succeeded except for kdevelop3.0.3-7 which depends on
libdb-4.1, which can't be found.
[Fix may have to wait for FTP version of SuSE 9.1]
---ooo---
Restoring my home directory.
This initially failed.
I have /home on a separate partition.
Before installing SuSE 9.1, I renamed /home to /homer (just in case).
During installation I used a different partition for /.
After installation I updated fstab to mount /homer and check,
then renamed /homer to /home in fstab and remounted.
This had all worked when I reinstalled SuSE 9.0, but failed for SuSE 9.1.
Why?
SuSE 9.1 has renumbered the users. My userid had been numbered 500 under SuSE
9.0 but SuSE 9.1 used 1000.
Consequently, when I switch back to the old /home partition it did not work,
because it did not recognize user 500 as me.
I tried fixing this by changing the number of my userid to 500.
This almost worked, but my previous logins had left files and directories
owned by user 1000 in various places, especially in /tmp.
I had to delete these before kdm and KDE would work correctly.
Now KDE 3.2.2 works well.
---ooo---
Fonts.
Anti-aliasing seems rougher than it was with SuSE 9.0.
This could be because I am not yet running the nVIDIA graphics driver.
Maybe sub-pixel hinting does not work in XFree86 4.3.99?
Testing the segfault problems with GluCat.
GluCat still segfaulted with kernel 2.6.4-54.5.
Updated to KOTD, which is
>uname -a
Linux linfinit 2.6.5-7.61cvs20040528150002-default #1 Fri May 28 19:04:18 UTC
2004 x86_64 x86_64 x86_64 GNU/Linux
GluCat has so far not segfaulted with this kernel.
---ooo---
32-bit software.
Maple will not install. Matlab will not run.
This is a showstopper, but I will try to fix it, if possible, rather than
revert to SuSE 9.0.
---ooo---
Other oddities so far.
Sometimes Konqueror response seems sluggish for no apparent reason, eg. CPU
usage is low.
---ooo---
Hardware:
AMD Athlon(tm) 64 Processor 3200+, on a Gigabyte GA-K8NNXP
motherboard, which uses the NVIDIA nForce3 150, plus an MSI FX5700-TD AGP
graphics card, which uses the NVIDIA GeForce FX 5700.
I'm totally amazed, totally ... who was the "smart" guy who figured out,
that on a 64bit system, the native libraries should be kept in a non-standard
directory lib64???? Every program I stumble upon, is choking on this ...
they all try to locate their "normal" libraries in /lib or /usr/lib and
expect them to be 64bit.
How come, this "smart" fellow didn't put non-native 32bit libraries as lib32
and 64bit native libraries where every-goddam-thing expects to find them?
Hello,
i hope you don't matter if i post my
next sound-related question here:
Every time i start KDE it tries to load
an ALSA sound driver even if there's
already running an other driver
(in my case OSS, i described in other
mails here why i can't use ALSA ...)
there allways appears a error message
related to this in ttyS10 .
I configured KDE in the controll center
to use OSS and it works but how could
i remove this ALSA dependencies or
whatever tries to load ALSA completely ?
I already removed all sound-
configurations with YaST and disabled
the ALSA modules before i installed
OSS, so i don't understand why there
are allways attempts to load it.
I don't know if this problem is related to my esound
problem i posted
here, maybe ?
best regards
Franz
SuSE 9.1 (amd64)
Mit schönen Grüßen von Yahoo! Mail - http://mail.yahoo.de
I'm unable to use Kdevelop with 9.1, first it doesn't find Qt libraries.
And if modified to find them, it doesn't compile programs that were fully
programmable with KDE 3.2 on SuSE 9.0.
Anybody notice, or fixed this yet?
Hi folks:
I would like to get a recent 2.6 kernel going on a 9.0 installation.
I am not sure if this would be harder than upgrading to 9.1, but I would
like to give it a try.
Do the RPMs indicated at ftp.suse.com/pub/people/aj/Kernel-AMD64/ work
with 9.0, or do I need to upgrade to 9.1? Any thoughts? If I wanted to
build from source, should I start with the .src or kernel-source rpm,
and work from there (to get all the SuSE patches)? Building kernels
with some other distro's RPMs can be a lesson in masochism, I just want
to make sure I am starting out right here.
Main purpose is to fix a customers issues with their 3ware card.
Thanks!
Joe
--
Joe Landman <landman(a)scalableinformatics.com>
Well, I had X Window hang on my AMD64 system.
Motherboard: MSI K8T800-FSIR
One SATA drive (boot, /) connected to Promise SATA connector
One ATA drive (not mounted by default)
The video card is an MSI card with NVidia GeForce FX 5200]
I am using the driver from NVidia (installed via YOU)
Just around the time of the crash, I am seeing a number
of messages like these:
Jun 2 08:28:22 mrburns kernel: Badness in pci_find_subsys at
drivers/pci/search.c:167
Jun 2 08:28:22 mrburns kernel:
Jun 2 08:28:22 mrburns kernel: Call Trace:<IRQ>
<ffffffff80209e43>{pci_find_subsys+83}
<ffffffff80209f10>{pci_find_slot+48}
Jun 2 08:28:22 mrburns kernel:
<ffffffffa0281830>{:nvidia:os_pci_init_handle+32}
<ffffffffa029d69e>{:nvidia:_nv000904rm+24}
Jun 2 08:28:22 mrburns kernel:
<ffffffffa0356eb4>{:nvidia:_nv002969rm+178}
<ffffffff80209e43>{pci_find_subsys+83}
Jun 2 08:28:22 mrburns kernel:
<ffffffffa03c51ff>{:nvidia:_nv001148rm+121}
<ffffffffa03f34b0>{:nvidia:_nv000711rm+966}
Jun 2 08:28:22 mrburns kernel:
<ffffffffa03735d7>{:nvidia:_nv001272rm+51}
<ffffffffa037274c>{:nvidia:_nv001217rm+28}
Jun 2 08:28:22 mrburns kernel:
<ffffffffa03f2eba>{:nvidia:_nv000704rm+46}
<ffffffffa0355892>{:nvidia:_nv002988rm+438}
Jun 2 08:28:22 mrburns kernel:
<ffffffffa03729e0>{:nvidia:_nv002809rm+16}
<ffffffffa03bb62b>{:nvidia:_nv001452rm+65}
Jun 2 08:28:22 mrburns kernel:
<ffffffffa03729b4>{:nvidia:_nv002811rm+16}
<ffffffffa0411ccd>{:nvidia:_nv005219rm+1861}
Jun 2 08:28:22 mrburns kernel:
<ffffffffa029e9ee>{:nvidia:_nv000719rm+62}
<ffffffffa029ec73>{:nvidia:_nv000858rm+103}
Jun 2 08:28:22 mrburns kernel:
<ffffffffa0353e25>{:nvidia:_nv002967rm+1907}
<ffffffffa0372344>{:nvidia:_nv001214rm+112}
Jun 2 08:28:22 mrburns kernel:
<ffffffffa029ee16>{:nvidia:_nv000884rm+16}
<ffffffffa03c140b>{:nvidia:_nv001134rm+723}
Jun 2 08:28:22 mrburns kernel:
<ffffffffa02a078e>{:nvidia:_nv005317rm+112}
<ffffffffa02a212d>{:nvidia:rm_isr_bh+9}
Jun 2 08:28:22 mrburns kernel:
<ffffffff801386fa>{tasklet_action+74} <ffffffff8013891b>{do_softirq+123}
Jun 2 08:28:22 mrburns kernel: <ffffffff80113391>{do_IRQ+257}
<ffffffff8011091d>{ret_from_intr+0}
Jun 2 08:28:22 mrburns kernel: <EOI>
<ffffffff80131a54>{thread_return+0} <ffffffff80110342>{system_call+26}
Jun 2 08:28:22 mrburns kernel:
<ffffffff801103a4>{system_call+124}
Jun 2 08:28:24 mrburns kernel: VFS: busy inodes on changed media.
Any ideas? hints? things to try to avoid this in the future?
--
----------------------------------------------------------------
Johannes Ullrich jullrich(a)euclidian.com
pgp key: http://johannes.homepc.org/PGPKEYS
contact: http://johannes.homepc.org/contact.htm
----------------------------------------------------------------
There are two kinds of system administrators:
The first kind solves problems with little shell scripts.
The second kind are the problem.
----------------------------------------------------------------
Slava,
I apologize, I misyped your name in my message, please forgive me.
Seán
On Wed, 02 Jun 2004 05:22:30 -0700 (PDT), sean-lynch(a)sean-lynch.com wrote:
>
> Slave,
> ...
Slave,
I have not used WSAD on my Athlon 64 yet, but I would advise 1 Gig minumum for
Websphere alone. At work our websphere only workstations use 1 gig. The
workstations with a Database running as well as WSAD get 2 or more gig.
These are pentium iv, 3.0 GHz boxes.
There are many ways to tune Websphere performance. Google is your friend. You
will find most advice for tuning the performance of applications you create with
webshere, but careful searching will give you sources for tuning WSAD studio.
Websphere will take as much memory as you give it. this is partly due to the
design of the IBM Java engine. As noted in this paper:
http://www.bmu-beratung.com/BMU-Download/pews.pdf
Here is an excerpt:
"For the IBM Java virtual machines (JVMs) the size of the Java
heap is controlled by the -Xmx parameter. So for example to
limit the Java heap to a maximum size of 128 MB, you would add
the following startup parameter
-Xmx128M.
If this parameter is not supplied, the default behavior for
an IBM JVM is to set the maximum Java heap size to one half
of the size of physical memory. If your computer had 512MB
of physical memory, the Java heap for each Java process could
grow to 256MB. Since the Java process requires memory for
things other than the Java heap, like for storing byte codes,
stack frames, native code, etc, the memory requirements for
the entire Java process could exceed 256MB. This other memory
is depicted as "Java Other" in the figures above, and it is
not affected by the -Xmx parameter."
So even going to 2 Gig, the JVM will take 1 Gig away, unless you limit it to
less.
Good Luck,
Seán
On Wed, 02 Jun 2004 13:41:33 +0200, slava zimine wrote:
>
> Hi to al & esp to Andreas Jaeger
>
> I'm seeking for advise to upgrade my workstation to amd64 platform.
>
> OS: suse 9.1
>
> main usage: (ejb/servlet swt gui java development )
>
>
> typical session load:
> Eclipse-gtk, or Websphere Dev Studio +
> Jboss or WebSphere App Server or Tomcat +
> oracle & mysql dbs +
> vmware sometimes (with XP loaded )
>
> Question: 1. what proc to choose ?
>
> 1 desktop Amd Athlon 64 3400 (2.2 GHz)
> or
> 2 x amd opteron 242 ( 1.6 Ghz)
>
>
> Are there java performance gains when using opterons which will
> justify +300$ price difference apart from a reason from sun that it will
> ship a jvm tuned specifically for amd64 opteron?
>
>
> 2. 1 or 2 Gig RAM Memory ?
> on my current ahtlon 1400 , 768 Meg Ram the WebSphere App
> Server takes almost 1.5 min to boot. I'd like to decrease this time
> .
>
>
> Thank you for you comments.
>
>
> slz
>
>
>
>
> --
> Check the List-Unsubscribe header to unsubscribe
> For additional commands, email: suse-amd64-help(a)suse.com
I'm not sure if my problem is related to the 64bit
version
or a general SuSE 9.1 issue, but as i'm using SuSE
AMD64
i'll post it here:
Related to my hardware i had to change to the
OpenSoundSystem (OSS)
to have sound because i had no chance to get ALSA
running.
The OSS is running with excellent sound quality but
now i tried to
use the Enlightened Sound Demon (esd) the first time
and something
very strange happened :
uruk:/home/paco # esd
- server format: sample rate = 48000 Hz
audio_alsa: no cards found!
uruk:/home/paco #
even if i define the sound-device in spawn_options in
the
esd Config files what shouldn't be necessary it fails:
uruk:/home/paco # esd
- using device /dev/dsp0
- server format: sample rate = 48000 Hz
ALSA lib pcm.c:1962:(snd_pcm_open_noupdate) Unknown
PCM /dev/dsp0
uruk:/home/paco #
i know that esd normally works with OSS, i had this
already running
with SuSE 6.0 (!) -
i have absolutely no idea why esd tries to open any
non existent
ALSA device - this seems to be a big riddle
as i removed all sound devices with YaST2 and then i
had disabled the
ALSA kernel modules using the OSS soundconf ...
Just for becoming sure that the problem isn't caused
by OSS in any
way i installed OSS on a other computer running
Mandrake 10
(as far is i know the program versions are very
similiar to SuSE 9.1)
the esd works out off the box ...
I'm really a bit confused now,
my only explanation for that all is that the SuSE 9.1
esd seems
to be broken in some way. Unfortunately i don't found
a patch for
it until nowwith YOU.
Am i the first having this problem ?
Please, could somebody tell me what is going on here ?
best regards
Franz
Mit schönen Grüßen von Yahoo! Mail - http://mail.yahoo.de