RE: [SUMMARY] NFS between Linux and Solaris

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



>  -----Original Message-----
>  From: redhat-list-admin@xxxxxxxxxx
>  [mailto:redhat-list-admin@xxxxxxxxxx]On Behalf Of Jason
Dixon
>  Sent: Wednesday, October 22, 2003 4:23 PM
>  To: Red Hat Mailing List
>  Subject: [SUMMARY] NFS between Linux and Solaris
>
>
>  Well, I wish I could say I have an authoritative answer
on
>  what caused
>  it, but I can no longer reproduce the problem.  I've
tried the same
>  thing with the following mount options, all of which work
fine:
>
>  vers=2,proto=udp
>  vers=3,proto=udp
>  vers=3
>  vers=3,proto=udp,noac
>  (no options)
>
>  At this point, I have to assume some other system and/or
>  network anomoly
>  was causing the problem.  If I can reproduce and resolve
the
>  symptoms,
>  I'll re-summarize.
>
>  -J.
>
>  On Wed, 2003-10-22 at 15:20, Jason Dixon wrote:
>  > Hi folks-
>  >
>  > My apologies if this is out there somewhere, but I've
>  googled this to
>  > death without finding a satisfactory answer.  I'm
>  attempting to tar copy
>  > a large repository (actually, the RHAS3.0 iso images)
from
>  a Linux NFS
>  > server to a Solaris NFS client.  At various intervals,
the transfer
>  > invariably dies with a "file not found" error.  The
cause
>  of this error
>  > can be explained by the sudden disappearance of
directories in the
>  > top-level of the exported share.
>  >
>  > Remounting the share causes the directories (and
>  everything recursive
>  > therein) to reappear, but the problem reoccurs the next
>  time I try the
>  > transfer.  I've tried a number of different flags, up
to
>  and including
>  > all of the following:
>  >
>  > vers=2,proto=udp,ro,noac
>  >
>  > Unfortunately, there's been no effect.  Has anyone run
>  into this?  Is it
>  > a known incompatibility between NFS implementations?
The
>  problem is
>  > easily reproducible.  The server is running Red Hat
>  Advanced Server 2.1
>  > Update 2 on a DL380.  The client is running Solaris 8
on
>  an E250.  Both
>  > servers, while on separate VLANs, are in the same
general
>  networking
>  > "vicinity".  Any ideas/solutions will be greatly
appreciated.
>  >
>  > TIA,
>  --
>  Jason Dixon, RHCE
>  DixonGroup Consulting
>  http://www.dixongroup.net
>

There is one thing wrong on redhat 7.1 that just bit me
yesterday.
maybe it is related to what is happening to you
might still be an unknown bug in the later releases.
At least worth a moments thought in terms of your situation.

I have Compaq proliants and redhat 7.1 (and later?) uses the
TLAN ethernet
driver on these systems.

This 7.1 system on an internal network that has not been
patched since the install
(yes, bad admin, bad admin, but its due for its REDHAT 9.0
upgrade)
deadlocked the ethernet interface after I started
both INBOUND and an OUTBOUND FTP sessions that pushed large
files
(like your NFS transfers)
into and off of the system. ISO images for Informix -
multiple 400K files.

We also had a large NFS transfer inbound from SUN solaris
2.6 server.
So the ethernet driver and software layer was very busy,
saturated in fact.

no sign of a resource problem on the box. netstat or
otherwise.

looks like I have a driver and/or TCP/IP software layer
bug to be patched. Sounds somewhat similiarlike what is
happening to you as well.
maybe it is not an NFS problem but an issue with the
underlying
network stack.

You are fine as long as traffic stays below some magic
threshold.
Push the interface past that point and weird stuff starts to
happen
until you get total failure usually requiring a reboot or
maybe
an interface reset via ifconfig.

The symptoms I got (see dmesg - /var/log/messages output if
it happens again)
were a kernel declaration  that the link had failed, it then
started
an endless loop of auto-negoitiation with the switch. Could
not seem to resync
even after i unplugged the cable for a few minutes. needed
to reboot to fix.

(I will have to lock those ports to 100/full someday soon
and turn off auto.)

You might be approaching that magic threshold with what you
are doing and causing
different symptoms (directory vanishing) than what i am
seeing here.



-- 
redhat-list mailing list
unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/redhat-list

[Index of Archives]     [CentOS]     [Kernel Development]     [PAM]     [Fedora Users]     [Red Hat Development]     [Big List of Linux Books]     [Linux Admin]     [Gimp]     [Asterisk PBX]     [Yosemite News]     [Red Hat Crash Utility]


  Powered by Linux