Hello,
I have tried 1.3.78 on my Suse 9.2 x86_64, with default kernel 2.6.8-24.11.
Openafs compiles fine, with sysname amd64_linux26, and I'm able to start afsd.
However, my first ls on our cell (not /afs) gives oops of bash:
Pid: 5202, comm: bash Tainted: PF U (2.6.8-24.11-default
Hello all,
I'm currently running a one server for fun cell on RedHat 9 with the
1.2.13 rpms. Is there any documentation about the upgrade procedure to
the latest 1.3.7x? What should I backup before trying this? Thanks for
any help.
-James
--
We are friends with those whose ideas are at the same
In message [EMAIL PROTECTED],Stephan Wiesand wri
tes:
With the UP kernel, I observe the write problem you described - with or
without memcache. Reading works in both cases. Shutting down the client
works as long as no write operations have been attempted, but is unusually
slow (several seconds
On Mon, 28 Feb 2005, chas williams - CONTRACTOR wrote:
In message [EMAIL PROTECTED],Stephan Wiesand wri
tes:
With the UP kernel, I observe the write problem you described - with or
without memcache. Reading works in both cases. Shutting down the client
works as long as no write operations have
I started seeing BreakDelayedCallbacks messages in the Filelog on one of
my file servers Friday afternoon. These messages have ceased since a
file server restart on Sunday morning. I searched for info on this
message, and all I could find is file server tuning may be necessary.
Any other ideas?
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hi, it's me again...
Now a question about time in the AFS cell.
My partner here told me, in the AFS cell should only one database-server enabled
ntp, all other should have ntp disabled.
Now I assume, if all clients has ntp enabled, all client has the
That is rather slow, though it can depend on various influences in your
environment. I am currently in the middle of a performance analysis on
the OpenAFS client, and can get 7-10MB/s on a 100Mbps ethernet (for
writes 7-8MB/s, for reads (non-cached data) about 9-10MB/s
(which is pretty much wire
In message [EMAIL PROTECTED],Stephan Wiesand wri
tes:
[EMAIL PROTECTED] ~]# grep -i preempt /lib/modules/2.6.9-5.0.3.EL/build/.config
# CONFIG_PREEMPT is not set
CONFIG_PREEMPT_VOLUNTARY=y
CONFIG_PREEMPT_VOLUNTARY means you get some preemption. just to test
this theory that its preempt, can you
NTP needs to run on all servers and workstations, use the real ntp not the
one bundled with AFS. Use the --nosettime switch to disable ntp in the AFS
server.
To keep your ISP happy, suggest pointing one or two AFS servers at 2 of the
the nearest Cicso routers and point the remainder of the local
On Mon, 28 Feb 2005, James Burns wrote:
Hello all,
I'm currently running a one server for fun cell on RedHat 9 with the
1.2.13 rpms. Is there any documentation about the upgrade procedure to
the latest 1.3.7x? What should I backup before trying this? Thanks for
any help.
Replace the binaries, bos
Hmm, it doesn't apply to a 1.3.79 tarball. There's nothing suitable for
the second hunk in that file. Applying the first hunk only makes inserting
the module fail:
libafs: module license 'unspecified' taints kernel.
libafs: no version for sys_close found: kernel tainted.
libafs: Unknown symbol
On Mon, 28 Feb 2005, Andrew Bacchi wrote:
I started seeing BreakDelayedCallbacks messages in the Filelog on one of
my file servers Friday afternoon. These messages have ceased since a
file server restart on Sunday morning. I searched for info on this
message, and all I could find is file server
Thanks to everyone who commented on this to the list. In the end it looks
like there is a stable configuration that works for both UP and MP
kernels. I have recompiled 2.6.9-5.0.3.EL and the corresponding libafs
module from scratch and all went well for the combination:
OpenAFS 1.3.79
On Mon, 28 Feb 2005, Andrei Maslennikov wrote:
To conclude, it seems that the problem had dissapeared between
2.6.9-5.EL/UP and 2.6.9-5.0.3.EL/UP kernels.
That's a bit scary, I wonder what changed?
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
Hello
I'm trying to setup Openafs 1.3.79 on Fedora Core 3. I've managed to
get authentication working correctly but when a user logs the
permissions on the files in his home directory are not set. The
directories in his home directory do have permissions set though. Does
anyone have any
Title: Force OpenAFS client to re-calculate Fileserver rankings
Is there a way to allow an OpenAFS Windows client 1.3.X to recalculate the ranking of Fileservers? Are rankings cached somewhere?
Any thoughts?
Roman Rozinov
Technology Support Analyst
Desktop Systems Technology,
On Monday, February 28, 2005 08:33:40 -0800 ted creedon
[EMAIL PROTECTED] wrote:
NTP needs to run on all servers and workstations, use the real ntp not the
one bundled with AFS. Use the --nosettime switch to disable ntp in the AFS
server.
This is somewhat misleading.
All servers and clients
All,
I a wondering if there is a way to artificially specify the /vicep quota
(or max size) when you have multiple namei fileserver partitions sharing
the same filesystem.
For example if I have a 150GB filesystem and want to put 5 /vicep
partitions /vicep[a-e] on it and wish to have no
Partitions don't have quotas... As far as I know there is no way to
accomplish what you are trying to do.
-- Nathan
Nathan Neulinger EMail: [EMAIL PROTECTED]
University of Missouri - Rolla Phone: (573)
Hi,
At one time, there was a problem with 1.3.76 such that mounts were not
created properly. I have a system which has this version of the
Windows OpenAFS client installed, and sure enough, does not work.
Installing a newer version, as expected also does not work. I'd like
to do a fresh
On Mon, 28 Feb 2005, Derrick J Brashear wrote:
On Mon, 28 Feb 2005, Andrei Maslennikov wrote:
To conclude, it seems that the problem had dissapeared between
2.6.9-5.EL/UP and 2.6.9-5.0.3.EL/UP kernels.
That's a bit scary, I wonder what changed?
Well, there was quite a pack of
On Mon, 28 Feb 2005, Paul Lussier wrote:
Is this sufficient to consider an installation of 1.3.77 a fresh
install and thereby not incur the wrath of this bug? Or is there
more to do than this?
That should be it. Reasonably modern OpenAFS doesn't leak registry keys
anymore, so if you had none
On Thu, 10 Feb 2005 15:51:19 -0500, Jack Tanner [EMAIL PROTECTED]
said:
Hi, just noticed that
http://grand.central.org/twiki/bin/view/AFSLore/WindowsLoopBackAdapter is
vandalized.
Dammit: I removed as much vandalization as I could last week; it looks
like it took two days for about 75 pages
On Mon, 28 Feb 2005, Joseph H Vilas wrote:
On Thu, 10 Feb 2005 15:51:19 -0500, Jack Tanner [EMAIL PROTECTED]
said:
Hi, just noticed that
http://grand.central.org/twiki/bin/view/AFSLore/WindowsLoopBackAdapter is
vandalized.
Dammit: I removed as much vandalization as I could last week; it looks
like
Excellent commentary.
Needs to be in the docs. Didn't realize there were 2 time slots involved.
Relying on a local timeserver works here but I'm not trying to sync outside
the local realm either.
tedc
-Original Message-
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
On Behalf Of
On Monday, February 28, 2005 17:45:08 -0500 Derrick J Brashear
[EMAIL PROTECTED] wrote:
So not that I have time to work on it now, and really we should get a
sysadmin for openafs to do this, but it would be nice to
1) require registration to edit the wiki
We did this a while ago for the
Well, FWIW, I've re-un-vandalized it. Only took about an hour and a
half this time -- I guess I'm getting better at it. ;)
Joe
--
Joseph H Vilas [EMAIL PROTECTED] 1-919-660-6902
Box 90132, Duke University, Durham, NC 27708-0132
I want to live in a society where people are intoxicated
with
27 matches
Mail list logo