August 2003 - openSUSE AMD64 - openSUSE Mailing Lists

RE: [suse-amd64] Tyan S2880 244 6GB
by Miller, Daniel J. 27 Aug '03

27 Aug '03

How severe was the "power problem?" I've had to re-install earlier Linux's after an accidental shutdown on a running system (wasn't on an Opteron system, but this sounds like a Linux filesystem issue rather than an AMD64 issue). I haven't seen it happen yet on SuSE 8.2, but I haven't seen many accidental power-offs on 8.2, either. It's one area in which NTFS seems to be a ahead of ext3 and reiserfs. Is the system on a UPS? Dan -----Original Message----- From: Santiago Flores [mailto:santi@mleads.com] Sent: Wednesday, August 27, 2003 11:05 AM To: suse-amd64(a)suse.com Subject: [suse-amd64] Tyan S2880 244 6GB Hello, I downloaded and installed Suse 8.2Beta. Straightforward install. No issues (other than having to turn off APIC - same for all other distros). The machine was up for about a week and a power problem required it to be shutdown. Now when I reboot the whole process hangs at the intialization of cpu #1. Any ideas? I booted with a rescue disk to checkout the disks and they were fine. I booted with a ClusterKnoppix CD that I had laying around and it worked fine, with both CPUs. Attached is the serial output of a boot with no APIC. Help please

1 0

RE: [suse-amd64] System freeze with NVIDIA 1.0-4180 on AMD64
by david.obrien＠amd.com 26 Aug '03

26 Aug '03

> Has anyone had any luck with the NVIDIA 1.0-4180 drivers on an AMD64 > system? Yes I have successfully used the driver with SLES 8. -- David

2 1

Re: [suse-amd64] Problem with strange CPU priority.
by Andreas Jaeger 23 Aug '03

23 Aug '03

On Sat, Aug 23, 2003 at 06:42:28PM +0200, Daniel Persson wrote: > Hi Andreas, > [...] > > You can bind processes to specific cpus via the /proc/interface or the > > numactl program. > > Ok, ill try that, but first, i downloaded the irqbalance program and have > it running on all machines. Before irqbalance /proc/irq/xx/smp_affinty > looked like this: > > > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > > And after it looked like this. That's fine. > 00000002 > 00000001 > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > 00000001 > 00000002 > 00000002 > ffffffffffffffff > 00000001 > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > ffffffffffffffff > 00000001 > (after running cat /proc/irq/*/smp_affinity > > Also a quick check with cat /proc/interrupts > CPU0 CPU1 > 0: 8578447 249920 IO-APIC-edge timer > 1: 15 0 IO-APIC-edge keyboard > 2: 0 0 XT-PIC cascade > 9: 794975248 0 IO-APIC-level acpi > 14: 35830 1 IO-APIC-edge ide0 > 15: 28 1 IO-APIC-edge ide1 > 19: 0 0 IO-APIC-level usb-ohci, usb-ohci > 24: 2288256 0 IO-APIC-level eth0 > NMI: 8575636 5437943 > LOC: 8829221 8829220 > ERR: 0 > MIS: 0 > > Show that CPU1 (this is right after running iqrbalance - so i guess that > it should change over time) have not handled near as much interruppts as > CPU0. Is my reading of this correct ? Is this normal ? The numbers are calculated since system start. The load is evenly distributed soon but not in the way that the same interrupte, e.g. for eth0 is handled by both cpu0 and cpu1 but in the way that CPU0 and CPU1 get in sum the same number of interrupts. So, cpu0 will now handle e.g. timer, acpi, eth0 and cpu1 will handle ide0, ide1 and timer. This also might give besser locality. > > > > > Are you using 2 2p boards in each case? Or a 4p one? > Both setups are with 2p boards. Ok. > > One question, im not near the machines right now, but, i they where > deliverd with all memory installed on CPU0, and no modules at CPU1, could > this generate this kind of problems ? It shouldn't ;-). For performance reasons I would change it. > (i could imagnie that it could) Andreas -- Andreas Jaeger, aj(a)suse.de, http://www.suse.de/~aj SuSE Linux AG, Deutschherrnstr. 15-19, 90429 Nürnberg, Germany GPG fingerprint = 93A3 365E CE47 B889 DF7F FED1 389A 563C C272 A126

2 1

[Fwd: Re: [suse-amd64] Problem with strange CPU priority.]
by Daniel Persson 23 Aug '03

23 Aug '03

Hi Andreas, > Replace it with a machine running SuSE ;-). Seriously, this shouldn't > be a problem. Nope, it should not, but this is the most obvius difference in the two setups, so i tought i mention it ;) > You can bind processes to specific cpus via the /proc/interface or the > numactl program. Ok, ill try that, but first, i downloaded the irqbalance program and have it running on all machines. Before irqbalance /proc/irq/xx/smp_affinty looked like this: ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff And after it looked like this. 00000002 00000001 ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff 00000001 00000002 00000002 ffffffffffffffff 00000001 ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff 00000001 (after running cat /proc/irq/*/smp_affinity Also a quick check with cat /proc/interrupts CPU0 CPU1 0: 8578447 249920 IO-APIC-edge timer 1: 15 0 IO-APIC-edge keyboard 2: 0 0 XT-PIC cascade 9: 794975248 0 IO-APIC-level acpi 14: 35830 1 IO-APIC-edge ide0 15: 28 1 IO-APIC-edge ide1 19: 0 0 IO-APIC-level usb-ohci, usb-ohci 24: 2288256 0 IO-APIC-level eth0 NMI: 8575636 5437943 LOC: 8829221 8829220 ERR: 0 MIS: 0 Show that CPU1 (this is right after running iqrbalance - so i guess that it should change over time) have not handled near as much interruppts as CPU0. Is my reading of this correct ? Is this normal ? > > Are you using 2 2p boards in each case? Or a 4p one? Both setups are with 2p boards. One question, im not near the machines right now, but, i they where deliverd with all memory installed on CPU0, and no modules at CPU1, could this generate this kind of problems ? (i could imagnie that it could) /Daniel

1 0

Problem with strange CPU priority.
by Daniel Persson 23 Aug '03

23 Aug '03

Hi list, I have a really strange problem, i have two different setups, one with Arima motherboards and one with Tyan motherboards. Both are running SLES8 and kernel k_smp-2.4.19-249 with all YAST2 updates. The code im running is Fluent 6.0.18 in parallell mode.(mpich) (CFD code) The problem is, that the load is very different when running in parallell, on the Arima based cluster, if running on 4CPUS (2dual machines) all cpus are occupied at 80-99% (wich should be considerd normal). However, running the exact same code and test case on the Tyan based machines(same kernel, network etc) then 2CPU is working hard and the other 2 are "watching" and this is one CPU per machine, ie one cpu in the dual machine is working, and CPU2 are idling. The result is that the same code and case are about 10times slower on the Tyan machines. (wich have the latest BIOS) The only(important?) difference in the setup is that the Tyan based test are connected to a RedHat 8.0 boxwith Dual Xeons. i.e auth and nfs are from this machine. But all computations are done on the Opteron machines. So, basically, my question are, could i have done some serious config mistake, or is there any problem with the kernel/Tyan bios wich casues one CPU to take alot more load than the other. Can i config the CPU routing somehow ? On the tyan machine, it also takes long time to just read the casefile.(ide disks, DMA and 32bitaccess enabled, same as Arima setup) Any ideas anyone ? /Daniel PS The same problems comes when using other codes aswell, such as LS-Dyna.

4 4

run-crons using an entire CPU
by Fred P. Arnold 22 Aug '03

22 Aug '03

Hello, I have a run-cron job which is executing the default scripts supplied with 8.2beta9 for AMD-64, on a Asus-based dual Opteron 240 system, which is using 100% of one of the CPUS. Top reports 6174 root 39 19 0 0 0 R 99.9 0.0 388:54.99 run-cron Any suggestion what's eating the time? -fred Frederick P. Arnold, Jr. NUIT, Northwestern U. f-arnold(a)northwestern.edu

2 1

glade-0.6.4 for SLES8-AMD64
by Stephen Williams 20 Aug '03

20 Aug '03

I had a need for this version of glade (which can co-exist with glade-2 which seems to be the new standard) so I used the src.rpm from the SuSE-8.2 Beta 9 disk. I've placed the rpms here: <ftp://ftp.icarus.com/pub/steve/x86-64/packages> There are some noarch.rpm prerequisites from the Beta9 disks that install cleanly, so get them from the disks. -- Steve Williams "The woods are lovely, dark and deep. steve at icarus.com But I have promises to keep, http://www.icarus.com and lines to code before I sleep, http://www.picturel.com And lines to code before I sleep."

1 0

System freeze with NVIDIA 1.0-4180 on AMD64
by Roderick Constance 19 Aug '03

19 Aug '03

Has anyone had any luck with the NVIDIA 1.0-4180 drivers on an AMD64 system? I have a SLES8-based sytem thas seems to have problems with an NVIDIA configuration. I can compile and load the 1.0-4180 NVIDIA_kernel and NVIDIA_GLX modules without problems. X starts up fine. But the system will lock up shortly after I start an OpenGL screensaver and move its window around. The X log and mtrrs look fine. Any suggestions? Thanks! -Roderick

1 0

AbiWord for SLES8
by Stephen Williams 19 Aug '03

19 Aug '03

Has anybody been able to build the latest AbiWord rpms for SLES8-amd64? I made a cursery attempt and got compiler errors even after I figured out the gnome-config paths. -- Steve Williams "The woods are lovely, dark and deep. steve at icarus.com But I have promises to keep, http://www.icarus.com and lines to code before I sleep, http://www.picturel.com And lines to code before I sleep."

2 6

rpm bugs on SLES8
by Stephen Williams 16 Aug '03

16 Aug '03

This is actually a distribution problem and not an AMD64 problem, but it is an AMD64 distribution, ... Anyhow, my source rpms for Icarus Verilog cause lots of warnings when I use "rpm --rebuild iverilog-0.7.20030815-0.src.rpm" to rebuild the package. Warnings like: I call /usr/sbin/Check... + /usr/sbin/Check rm: cannot remove `/usr/share/info/dir.bak.gz': Permission denied gzip: /usr/share/info/dir.bak.gz: Permission denied -rw-r--r-- 1 root root 503 2003-07-17 07:06 /usr/share/info/dir.bak.gz rm: cannot remove `/usr/share/man/man1/gimp-1.2.1.gz': Permission denied gzip: /usr/share/man/man1/gimp-1.2.1.gz: Permission denied -rw-r--r-- 1 root root 5318 2003-07-09 06:53 /usr/share/man/man1/gimp-1.2.1.gz In fact, it appears that the subcommands are not using the BUILD_ROOT that is there to allow for non-root rebuilds of packages. It may be that an old version of rpm is being used. The srpms in question is here: <ftp://ftp.icarus.com/pub/eda/verilog/snapshots/precompiled/verilog-0.7.2003…> -- Steve Williams "The woods are lovely, dark and deep. steve at icarus.com But I have promises to keep, http://www.icarus.com and lines to code before I sleep, http://www.picturel.com And lines to code before I sleep."

2 6