March 2004 - openSUSE AMD64 - openSUSE Mailing Lists

RE: [suse-amd64] effective memory in Opteron system
by daniel.kidger＠quadrics.com 25 Mar '04

25 Mar '04

For information , we found that one particular BIOS for Quartets would 'ignore' one or more memory slots. Any DIMMS there were not discoveered by the BIOS. The actual slots affected seemed arbitrarty - the same node would always have teh same slot(s) affected - but different identical nodes had different problem DIMM slots. When you added that 4th DIMM you could try using a different slot to see if it is any better. (yes I know about paired memory and 128-bit access but this is a test) A BIOS rev fixed this for us- you do not say what you BIOS revision is. /usr/sbin/hwinfo is your frined here - it not only tell you the BIOS version , but also the detected DIMMS by slot number and also the A20 physical memory map (hence spot the memory holes) Finally IMHO, the best way forward if Linux is not seeing all the memory you paid for (but the BIOS is seeing it) is to have say a 2GB hole from 2-4GB and push the rest of the memory up. AMD/Celestica have a BIOS that does this that we use. Yours, Daniel. -------------------------------------------------------------- Dr. Dan Kidger, Quadrics Ltd. daniel.kidger(a)quadrics.com One Bridewell St., Bristol, BS1 2AA, UK 0117 915 5505 ----------------------- www.quadrics.com -------------------- > -----Original Message----- > From: Andi Kleen [mailto:ak@suse.de] > Sent: 25 March 2004 01:39 > To: ascotti(a)email.unc.edu > Cc: suse-amd64(a)suse.com > Subject: Re: [suse-amd64] effective memory in Opteron system > > > On Wed, 24 Mar 2004 20:21:58 -0500 > ascotti(a)email.unc.edu wrote: > > > Hi, > > > > I have recently purchased a quad-opteron. The MB is a > Quartet manufactured > > by Celestica. The system has 8 GB of RAM (8 dimms, 2 GB per > node). The > > puzzling thing is that only 7.5 GB show up, both during the > BIOS startup, > > and by the OS (SLES 8). Playing around with the dimms, I > have found that up > > to 3 GB everything looks normal. Adding a fourth dimm to > the system results > > in an effective memory of 3.5 GB. Adding extra modules > increase the count > > by 1GB as expected. The dimms are good, and nothing changes > by switching > > dimms or populating the nodes differently. The vendor tells > me that this is > > a known "feature" of the AMD chipset, even for Athlons. I > looked up the > > thing, and found nothing. I am curious to see if this is > just a problem > > with the Opteron chipset, or a more general problem. 0.5 GB > seems a lot to > > disappear... > > It depends on the BIOS. The hardware can only map a single > DIMM to an continuous > address. Your BIOS decides to map the first three DIMMs at > 0,1GB,2GB. Then > there is the PCI memory hole directly below 4GB (PCI memory > mappings, AGP > aperture, IOMMU etc., can be several hundred MB). Your BIOS > could chose to leave > 3GB-4GB free and only put the IO mappings there and put the > next DIMM at 4GB, > but it chose instead to map the DIMM at 3GB and so part of > its memory is lost > to the IO hole. > > There may be subtle reasons why it does that. > > Complain to your BIOS vendor about this, there is nothing > Linux has to > do with it. > > -Andi > > > -- > Check the List-Unsubscribe header to unsubscribe > For additional commands, email: suse-amd64-help(a)suse.com > >

1 0

Network Configuration Scripts
by John Craig 25 Mar '04

25 Mar '04

Hi, Can anybody give me an idea about how to properly set up a server to be configured for ethernet trunking at boot time? Our file server is a dual Opteron running SuSE 9.0 for AMD64, and has dual gigabit ethernet cards, and I have them both connected to the switch (with trunking enabled for those ports). I am doing this with a kluge, but have not had the time or patience to figure out how to do it better. The commands I use are: ifconfig eth0 down ifconfig eth1 down ifconfig bond0 192.168.0.100 ifenslave bond0 eth0 ifenslave bond0 eth1 The modules.conf file has: alias bond0 bonding alias eth0 bcm5700 alias eth1 bcm5700 In YaST, I configured eth0 and eth1 with dummy ipnumbers, just to make sure that the driver loads and the interfaces exist, that's the reason why they have to be stopped before bringing up bond0. I added "alias bond0 bonding" manually to modules.conf. Is anybody else out there using ethernet trunking (bonding), and how to you set it up? Thanks, John Craig

3 5

RPM of Samba 3.0?
by Quy NGUYEN DAI 25 Mar '04

25 Mar '04

Hi, Is there somewhere a RPM of samba v3.0 for AMD64 ? I tried rebuild RPM from src.rpm (http://ftp.sernet.de/pub/samba) but ot not works: ... Compiling lib/md5.c In file included from lib/md5.c:23: /usr/src/packages/BUILD/samba-3.0.2a/heimdal/include/md5.h:55: error: conflicting types for `MD5_CTX' include/md5.h:22: error: previous declaration of `MD5_CTX' make: *** [lib/md5.o] Error 1 error: Bad exit status from /var/tmp/rpm-tmp.79375 (%build) RPM build errors: Bad exit status from /var/tmp/rpm-tmp.79375 (%build) ... Any ideas for help? Thanks in advance, -- Quy NGUYEN DAI

1 0

LDAP client
by masken＠gentoo.se 25 Mar '04

25 Mar '04

Hi, I have som problems with automount on suse-9 for amd64. I have /etc/nsswitch.conf's automount line set to ldap only. (automount: ldap). And the /etc/ldap.conf and /etc/openldap/ldap.conf files are corectly setup and i can authenticate properly with ldap. This is what i get in messages. Mar 24 22:46:35 hsgi16 automount[1625]: cannot open lookup module nisMapName (/usr/lib64/autofs//lookup_nisMapName.so: cannot open shared object file: No such file or directory) Mar 24 22:46:49 hsgi16 automount[1628]: cannot open lookup module ldap:auto_homes (/usr/lib64/autofs//lookup_ldap:auto_homes.so: cannot open shared object file: No such file or directory) But when i manually start automount with, /usr/sbin/automount /home ldap nisMapName=auto_home,dc=my,dc=company,dc=org it works like a charm. Anyone? Thank you. ///Mathias

1 0

RE: [suse-amd64] 9.1 Pro for AMD64 and SATA/IDE RAID ?
by Hulslander, Ryan 25 Mar '04

25 Mar '04

> -----Original Message----- > From: Andreas Jaeger [mailto:aj@suse.de] > > Lastly, if the drivers are included and they do load/work during a > > bare-metal install, are they still purely experimental at > this point? > > SATA is still quite early. > Then I assume if YAST recognizes the SATA drive I'm installing the OS on to, it'll really just use some sort of experimental/beta driver to interface with the drive? Okay, thanks. That's what I needed to know. I guess I'll stick with IDE for the moment then when I install 9.1 Ryan

2 1

SLES 8 / Service Packs, unixODBC
by Werner, Sascha 25 Mar '04

25 Mar '04

Hi! Sorry for the stupid questions... Where can I find the Service Pack 3 for United Linux 1.0? On the SuSE websites, I'm just able to locate SP2. And where can I get a rpm of unixODBC 2.2.8 for SLES 8? Thanks in advance, Sascha

2 1

Re: [suse-amd64] yast online update
by einar 25 Mar '04

25 Mar '04

Hello, I played around with Yast to see if I could change the directory where Yast is trying to find updates but without success (I even reinstalled yast). I have the suspicion that the problem is that I installed from the 9.0 patch CD (for the SATA drivers). YOU claims that my product is the "SuSE Linux PatchCD". Somehow if you boot from the 9.0 patch CD the Patch CD inscribes itself as the product and not the 9.0 DVD you have to use afterwards during the install. Does anyone know how I can change this? regards, einar

1 0

suse 9.0 amd64 panics kernel on zv5000z bad rip
by bob sandefur 24 Mar '04

24 Mar '04

Hi- I tried to install 9.0 for amd64 on my new zv5000z from hp but I get an bad rip kernel panic right out of chute however Fedora for amd64 installs and works Mandrake 9.2 and 10 for amd64 install and work The install cd is ok because it worked on my opteron 144 last year Any similar problems or ideas Thanx

1 0

Suse 9.1 and will it have pppoE instalation
by Benjamin Hamilton 24 Mar '04

24 Mar '04

Could there be an option for pppoE ftp installation? I know its a ways off but I think the bootdisk should be graphical and use yast2 am I able know to use pppoE to intall over ftp I think not but I could be mistaken. Sure it is a beautifully simple program but it is not easily upgraded and I don't want to have to do it myself. But I would know where to look but I'd have to recompile and there just isn't enough need for it I guess. _________________________________________________________________ Find a broadband plan that fits. Great local deals on high-speed Internet access. https://broadband.msn.com/?pgmarket=en-us/go/onm00200360ave/direct/01/

1 0

install goes ok, nothing after reboot
by James Boorn 24 Mar '04

24 Mar '04

I am trying to install Suse Linux Professional 9.0 for amd64 on a new Appro 2128Hs server. It is a dual Opteron 246 system, Tyan S2881 motherboard. No dvd, so I do an nfs install (on to scsi disk (seagate disk, adaptec aic79xx driver, custom partition scheme) which appears to go smoothly. Everything I need is detected, doing a 'Default system' for the package selection. However, upon the reboot the boot loader doesn't appear to want to load (tried grub and lilo). All I get is a blinking under line cursor. Any ideas? Thank You, James

1 0