Anybody have any problems with amd on 2.2_beta

Anybody have any problems with amd on 2.2_beta

Post by Marcelino Enrique » Wed, 29 Jan 1997 04:00:00

I was wondering if anybody out there is using amd to mount all
of there disks that are on seperate servers.

Here is my problem. I start amd, and by the way I have set the nfs
client option in /etc/sysconfig. Everytime I access a directory that
is mounted from an alpha running osf/1 ver 3.2 it hangs. Nothing I do
to the shell prompt will ever cause it to let go. The machine remains
active just the prompt never is released. I am not sure what is causing
this, but non of our other systems seem to exhibit this and we are
currently running solaris 2.5.1 x86, hpux 9.07, and irix 5.3 as well
as some old dec machines running ultrix. I am sorry that I can't supply
some of the auto maps or even the sysconfig. The machines that were
originally up and running have been turned off temporarily, but I
thought I'd ask anyway.



1. Problems with kernel 2.2.x and amd/NFS

Hi there!
We are using Linux2.0.36/amdpl102 in combination with HPUX9.0.7/10.20.
Due to hardware upgrades not supported by 2.0kernel we are forced
to use 2.2 kernel.

Automounting hp directories gives a seemingly empty directory, but
executing shell script or copying files from or to the automounted
directory works. Recompiling the amd on the 2.2 machines ameliorates
the situation somewhat, directories containing no more than 20-30
entries are listed correctly, dir's with more entries are not
listed completely; there's no error message but the /var/adm/messages
shows entries like 'NFS: server bla, readdir reply truncated'.
Changing rsize/wsize in the amd config changes nothing, this seems
to be a limit in the kernel nfs itself, the kernel level code simply
bails out from a loop when a certain size limit for output is reached.
(fs/nfs/nfs2xdr.c if it's of any interest)

However, I don't see any possibility/changeable parameter to increase
this limit in the kernel code and simply omitting the bailout seems
to dangerous to me.

Has anyone met the same problem/does know if there is a workaround?
PS: autofs is not an option since we need an unknown number of unknown
directory names on the remote. autofs-3.1.3 which I tried has no capa-
bility of this type even with --submount.

Thanks for the help
Remember: The Game of Life has no 'restore' Button!
Busted once, Busted forever.

2. Answerbook

3. Linux 2.2.x wishlist: kernel automounter; AMD problem

4. Paradise SVGA card under SVGAlib?

5. Any of 2.2.x not have CPUID problem with AMD CPUs?

6. Profitability of Linux being a challenge

7. Anybody else having problems with missing data using AHA-2940???

8. Netscape 4 crashes AIX 4.2.1

9. Question: Is anybody having problem with Adaptec 2842 ?

10. Anybody having problems with alt.os.linux and FA?

11. Anybody having stability problems w/Redhat 6.0?

12. Anybody having problems with Matrox Mystique and X server?

13. Upgraded from 2.2-GAMMA to 2.2-RELEASE: problems...