On Thursday, 3. January 2002 11:00, lv426@rinx.homeip.net wrote:

Kurt Seifried wrote:

...

simply download samba source rpm, samba tarball, edit samba spec file, copy in tarball, voila. rpm makes it easy to build your own upgrade packages.

Kurt Seifried, kurt@seifried.org A15B BEE5 B391 B9AD B0EF AEB0 AD63 0B4E AD56 E574 http://www.seifried.org/security/

----- Original Message ----- From: "Christoph Egger" To: Sent: Wednesday, January 02, 2002 6:19 AM Subject: [suse-security] kernel panic under high load

...

Hello!

I use SuSE 7.3 on a file server with Samba 2.2.1a (shipped with SuSE 7.3). But under high load, the SuSE Kernel panics.

I am now downloading kernel 2.4.17 to compile and install myself.

And I want to update Samba to 2.2.2, because of some oplock fixes. Is a ready SuSE rpm somewhere available? If not, which configure options have I to use in order to overwrite the

old

...

binaries and that the binaries searches smb.conf in /etc/samba?

-- CU, Christoph

Hi Christoph,

this should do: <--snap

./configure --prefix=/usr/lib/samba --bindir=/usr/bin --sbindir=/usr/sbin --libdir=/usr/lib/samba --sysconfdir=/etc/samba --mandir=/usr/share/man --with-privatedir=/etc/samba --with-lockdir=/var/lock/samba --with-configdir=/etc/samba --with-swatdir=/usr/lib/samba/swat --with-smbwrapper --with-dce-dfs --with-automount --with-smbmount --with-msdfs --with-libsmbclient --with-acl-support --with-winbind --with-vfs --with-quotas

snap --->

TNX, a lot! The fileserver runs very well now! Kernel 2.4.17 + Samba 2.2.2 is a very stable combination! I stresstested it by coyping and moving lots of data around (about 30 GB) and the server runs and runs and runs... :) Another problem: NFS! When I try to mount a directory via nfs, the fileserver prints in /var/log/messages: ---------------------------------------- Jan 3 13:59:45 fileserver mountd[18775]: NFS mount of /home/install attempted from 192.168.2.91 Jan 3 13:59:45 fileserver mountd[18775]: /home/install has been mounted by 192.168.2.91 on the client side, the mount command needs a loooong time to exit. When I make a "ps axf | grep mount", then I get: ------------------------------------------ 1911 pts/5 D 0:00 \_ mount -t nfs 192.168.2.70:/public/install ./install Does anyone know, what goes wrong? -- CU, Christoph

On Thursday, 3. January 2002 16:40, markus@gaugusch.at wrote: [my own problem description]

You need to start portmap (rcportmap start) on client side (enable it in rc.config to have it running after booting).

TNX a lot. Works ok! -- CU, Christoph

lockd? quotad? portmap is useful. Kurt Seifried, kurt@seifried.org A15B BEE5 B391 B9AD B0EF AEB0 AD63 0B4E AD56 E574 http://www.seifried.org/security/ ----- Original Message ----- From: "Peter Wiersig" To: Sent: Friday, January 04, 2002 1:24 AM Subject: Re: [suse-security] kernel panic under high load

Am Donnerstag, 3. Januar 2002 16:36 schrieb Markus Gaugusch:

...

[lot of stuff deleted - please people: learn to quote!!]

Who said the following paragraph? You should not delete that.

...

...

Another problem: NFS! When I try to mount a directory via nfs, the fileserver [...] on the client side, the mount command needs a loooong time to exit. When I

You need to start portmap (rcportmap start) on client side (enable it in rc.config to have it running after booting).

Hi Markus,

can you or somebody on this list explain to me why nfs-mounts need portmap on the client?

Or is there some documentation which I oversaw?

I have "Building Internet Firewalls, 2nd Ed." in which RPC-based services such as nfs are described as this:

server registers to portmapper ... client asks portmapper "where is nfs" portmapper responds "nfs ist on port <abcd>" ... client requests server

I have googled a bit but never found a hint why portmap must be running on the client.

If I don't have portmap the mount succeeds after a long time and has no further delays when accessing files.

If I start portmap, mount the share, and kill portmap afterwards everything seems to do work.

Can someone enlighten me, please?

Peter

-- To unsubscribe, e-mail: suse-security-unsubscribe@suse.com For additional commands, e-mail: suse-security-help@suse.com

On Thursday, 3. January 2002 11:27, jsa@pen.homeip.net wrote:

On Wednesday 02 January 2002 04:19 am, Christoph Egger wrote:

...

Hello!

I use SuSE 7.3 on a file server with Samba 2.2.1a (shipped with SuSE 7.3). But under high load, the SuSE Kernel panics.

I am now downloading kernel 2.4.17 to compile and install myself.

I noticed no one addressed the subject of whether the kernal panic is samba related or kernel related. What lead you to believe it was either?

Well, the keyboard LED's were blinking, no disk activity, no network activity... I couldn't login via network (ssh/telnet), nor directly from the terminal. A hard reset was a must. As of this writing I am running 2.4.17 now. I am going to update Samba (TNX for the answers, btw) and will retest, if that fixed the problems.

Or are you trying a shotgun approach.

(Interested, because I've just installed two such server in the last two weeks and have seen no panics even when dumping tons of data onto the servers drives thru samba).

Which kernel do you run? Do you use the SuSE Kernel coming with SuSE 7.3? Which version of Samba do you use? -- CU, Christoph