[OpenAFS] OpenAFS 1.4.2 on WindowsXP
Hi, I have a strange behavior with the new release of openafs. I tested openafs-1.4.2.-beta1, because of problems with some office tools. Now I want to update all the computers in our production environment from Version 1.4.1 on 1.4.2. After the installation I get no token during login. After login I could obtain a token manually. When I step back to 1.4.1 or 1.4.2-beta1, there are no problems. I set the option to debug the integrated login. Sometimes I don't see messages about AFS-Logon and sometimes the message appears "Authentication server was unavailable". Are there any changes? Or does anyone have an suggestions? Regards, Lars -- Dipl.-Inf. Lars Richter Pro Design Electronic GmbH Hochheimer Straße 47, 99094 Erfurt phone: +49 361 55038-25 fax: +49 361 78930-80 www.prodesigncad.de ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AFS client suddenly extremely slow
Today at 15:53, Jeffrey Hutzelman <[EMAIL PROTECTED]> wrote: I believe the auto-tuning made it into 1.4.1; it's definitely in 1.4.2. However, that doesn't help you if you insist on passing explicit values on the command line, as is done by the startup scripts included with several of the binary packages. -- Jeff Any chance someone could rattle off the command line options that are not needed if we want to use the auto-tuning? We currently use something like -stat 2800 -dcache 2400 -daemons 5 -volumes 400 for the afsd options on or Solaris 10 clients running 1.4.2. If it's in a release notes somewhere that I missed, feel free to point me in that direction, too. Thanks! --Kris -- Thomas Kris Kasner Qualcomm Inc. 5775 Morehouse Drive San Diego, CA 92121 (858)658-4932 ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AFS and PhotoShop interaction?
Jeffrey Altman wrote: I'd suggest trying the 1.5.10 client. I have a couple people with 1.5.10 clients installed, and they experience the same behavior. I am now borrowing a WinXP box and it's sitting on my desk. I experience the same behavior. Follow the directions in the release notes for capturing log data with the SysInternals FileMon and DbgView tools. FileMon will show what operations the applications are requesting. DbgView will capture what OpenAFS is doing in response. Set both tools to capture with millisecond clock time and the logs can be examined in lock step. Send a bug report with the logs to [EMAIL PROTECTED] Jeffrey Altman Ok... I've been using Filemon, and it's giving very enlightening information. I'm a bit loathe to open an afs bug report, because I'm now convinced it has little to do with AFS, except that the paths are kinda deep because of the way the tree is set up, which makes all those OPEN DIR calls take longer as it traverses, and maybe there's an issue with the amount of time it takes for QUERY INFORMATION to complete. Is there an equivalent of "fakestat" for the windows client? I think that's only for things crossing cell boundaries... but anyway... in case anyone is interested... I was using Photoshop CS2 9.0. I'm upgrading to 9.0.2 now just in case. What appears to be happening is that when I want to open a file, Photoshop decides that it must get a status on every file I've opened since I started the program... twice So, for example, if I opened files A.jpg, B.jpg and C.jpg in C:\TEMP, and then I open D, E and F.jpg on my desktop, and then I open H:\mycell.com\assets\images\*.jpg by dragging, say 10 jpg images in then I want to open a file C:\myimg.jpg. In Filemon, I see photoshop open and read myimg.jpg. And then, stat a lot of files... It opens each piece of the directory path to A.jpg, and then does a QUERY INFORMATION on A.jpg... then each piece of the directory path to B.jpg, then QUERY INFORMATION on B.jpg, etc. This goes for every file I've opened recently, so it hits AFS as well, and afsd_service.exe uses up about 50% of the cpu. Then, it starts all over again my doing a bunch of alternate READ and QUERY INFORMATION calls on myimg.jpg, and then repeats all the other stuff starting with A.jpg and does the QUERY INFORMATION calls for all of the files again. It looks like it does a QI on the last 30 files opened. If I open 30 files off the hard drive, then open more files, it works ok. If I open 30 files out of AFS, it slows down with each open. If there is a way to tune the AFS client to speed up the time it takes to return the information Photoshop is requesting, please let me know what that is. I'm going to open a ticket with Photoshop support to see if there is a way to get it to stop doing what it's doing. The open file history survives reboots. So the only way to clear it seems to be to open 30 small local files. I am NOT going to tell my users that. ;-) Thanks to everyone for the help and suggestions. Let me know any tuning suggestions to speed up the QUERY INFORMATION calls. Mike P. P.S. Jeffrey-- if you still want filemon logs for any reason, let me know. Seemed like it would just be chaff in the queue at this point... ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AFS and PhotoShop interaction?
Mike Polek <[EMAIL PROTECTED]> wrote: Christopher D. Clausen wrote: Mike Polek <[EMAIL PROTECTED]> wrote: information and would like to know if anyone else has had a similar experience. This is on Windows XP using 1.4.2a-DEBUG client. The fileserver is running Fedora Core 5 and OpenAFS 1.4.2 (new machine). Similar behavior using 1.4.0 and 1.4.1 windows clients and 1.4.1 fileserver. I'd suggest trying the 1.5.10 client. I have a couple people with 1.5.10 clients installed, and they experience the same behavior. I am now borrowing a WinXP box and it's sitting on my desk. I experience the same behavior. Hmm... could it be network trouble? Are you sure the duplex setting is correct on all intermediate equipment? Do you have some switches that are doing stupids things (we had some that were doing cut-through switching instead of store-and-forward and that was causing problems)? Also, run netstat -s before you start a session and compare it to afterwards and see if you are seeing an abnormally high number of retransmits or other problems. How big are the files? Are they larger than the AFS cache on the client? Files are about 2-4 MB. Cache is 98MB (256MB on some computers). Hmm... That shouldn't be a problem. Can you try increasing the chunksize to see if it helps? I think the default is 256K. If you are working with larger files a larger chunksize may help you out. fs setcrypt off seems to have no effect. What does have an effect is whether or not AFS is running... even if the files are on the local hard drive. Really? That should most definately NOT be a problem... Photoshop isn't trying to use some file in AFS as swap of temp space is it? You can use filemon (from sysinternals) to monitor file I/O and see if any network I/O occurs when running photoshop with local files. Oh... hmm... Can you try disabling the AFS context menus? I use: http://www.nirsoft.net/utils/shexview.html to disable the various AFS bits for testing certain things. Now this shouldn't cause problems, but if you say local accesses are affected, this is one of the few things I can think of that might be the cause.
Re: [OpenAFS] Compiling afs module on sparc64
Bastian <[EMAIL PROTECTED]> writes:
> I don't know if this is of much help, but I recently compiled the
> openafs module 1.4.2 on Etch (2.6.17) i386 without any problem. However:
> on loading the module dmesg complaints about the license ('taints
> kernel'). It seems the license of openafs does not look GPL enough to
> linux. The module loads though.
Yup, that's because it's not GPL (it's released under the IBM Public
License).
However, normally it will build anyway. It sounds like there's some sort
of additional restriction on sparc64 that's not present on x86 or amd64
(at least) in the kernel.
> It looks to me your problem is some kind of 'refusal' to build openafs,
> rather than a technical issue. Could there be some policy in effect on
> your system?
Usually this policy is set in the kernel source, I'm afraid.
--
Russ Allbery ([EMAIL PROTECTED])
Re: [OpenAFS] Compiling afs module on sparc64
Gunnar Krull schreef:
Hi,
I'm trying to compile the openafs module without success on a dual sparc64 Sun
server running Debian Etch.
With version 1.4.2 and 1.5.10 a make shows me this error message:
echo 'char* AFSVersion = "openafs 1.5.8"; '
>>/tmp/openafs/openafs-1.5.10/openafs-1.5.10/src/libafs/MODLOAD-2.6.18-1-sparc64-smp-MP/AFS_component_version_number.c
CC [M]
/tmp/openafs/openafs-1.5.10/openafs-1.5.10/src/libafs/MODLOAD-2.6.18-1-sparc64-smp-MP/AFS_component_version_number.o
LD [M]
/tmp/openafs/openafs-1.5.10/openafs-1.5.10/src/libafs/MODLOAD-2.6.18-1-sparc64-smp-MP/libafs.o
LD [M]
/tmp/openafs/openafs-1.5.10/openafs-1.5.10/src/libafs/MODLOAD-2.6.18-1-sparc64-smp-MP/afspag.o
Building modules, stage 2.
MODPOST
FATAL: modpost: GPL-incompatible module libafs.ko uses GPL-only symbol
'sys_call_table'
The same result with smp kernel 2.6.17 and 2.6.18.
Is there a way to compile the afs module on this system?
Thanks for any suggestions!
Gunnar
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info
I don't know if this is of much help, but I recently compiled the
openafs module 1.4.2 on Etch (2.6.17) i386 without any problem. However:
on loading the module dmesg complaints about the license ('taints
kernel'). It seems the license of openafs does not look GPL enough to
linux. The module loads though.
It looks to me your problem is some kind of 'refusal' to build openafs,
rather than a technical issue.
Could there be some policy in effect on your system?
Bastian
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info
[OpenAFS] Re: [OpenAFS-win32-devel] OpenAFS vs Vista
Jeff, I recall seeting the IFS working (by some definition of "working") on Eric JW's workstation in 2005. What specifically are the missing features that must be implemented to complete the work? Matt Jeffrey Altman wrote: The long term solution is clear. The OpenAFS community needs to move away from the SMB Gateway model and replace it with a native Network Provider / File System Filter model. Completing this work in such a way that it is portable across Windows 2000 SP4 through Vista and works on both 32-bit and 64-bit operating systems is going to be time consuming and expensive. I hired a Windows file system and kernel driver developer to review the work that has already been done and help design an architecture that can support all of the required platforms for the next ten years. We believe that this work can be accomplished over a period of ten months provided that the necessary financial resources ($250,000 - $300,000) can be acquired. The long term benefits of getting rid of the SMB gateway are: (1) the use of the Microsoft Loopback Adapter will no longer be required (2) all of the problems associated with the CIFS client timeout issues will be removed. this will not only improve performance but will solve all of the "delayed write" and "network path no longer available" errors. (3) performance gains. There is significant time delays and CPU load added to each transaction as a result of the CIFS client and SMB server both involving the SMB/NetBIOS/TCP/IP stack. If your organization is in a position to assist us in financing this work, please contact me off-list. Jeffrey Altman OpenAFS Gatekeeper/Elder ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] LiveCD for openafs disaster recovery scenarios ?
I have a suggestion... if you have another machine with the same O/S, export the /usr partition via NFS readonly. Change your fstab on the damaged host to mount /usr from the NFS server host. It should work, similar to the way diskless booting works. Once the machine is up and running, you could just move the volumes off. Make sure the other machine has the same version of openafs installed, or at least make sure your files are consistent between what's on /usr and what's in your / partition. Boot off a rescue CD to update your fstab file. Good luck! Mike Madhusudan Singh wrote: Hello At least one (and possibly two) hard disks on the machine (has RAID 5 - rebuild did not help too much) that runs our openafs cell failed today. This is a Debian machine. Partition /usr is down and attempts to mount it fail with bad superblock messages. As a result, all network and many other system commands (awk, etc.) are inaccessible. Now, is there a live CD that can be used to run the machine so that the users can at least copy over their data ? I think a reinstall is indicated. The two partitions housing /vicepa and /vicepb are not affected. Is there a way to retrieve data directly from the hard disks ? Thanks. ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info -- Michael Polek Pictage, Inc. ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] Re: [OpenAFS-win32-devel] OpenAFS vs Vista
Dan Pritts wrote: > On Tue, Nov 07, 2006 at 08:53:28AM -0800, Jeffrey Altman wrote: >> I want to provide you with an update on OpenAFS compatibility and >> Windows Vista. A couple of months ago a bug was discovered in the >> Vista NetBIOS stack that would cause OpenAFS to crash anytime the >> operating system was placed into "Standby" mode. Microsoft took > > does the bug also exist when you go into "hibernate" mode? > or just "standby"? > > tnx > danno I believe Vista removes hibernate mode but it wouldn't matter. The bug would be present anytime the machine power's down the network interfaces. Jeffrey Altman smime.p7s Description: S/MIME Cryptographic Signature
[OpenAFS] Re: [OpenAFS-win32-devel] OpenAFS vs Vista
On Tue, Nov 07, 2006 at 08:53:28AM -0800, Jeffrey Altman wrote: > I want to provide you with an update on OpenAFS compatibility and > Windows Vista. A couple of months ago a bug was discovered in the > Vista NetBIOS stack that would cause OpenAFS to crash anytime the > operating system was placed into "Standby" mode. Microsoft took does the bug also exist when you go into "hibernate" mode? or just "standby"? tnx danno ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
[OpenAFS] OpenAFS vs Vista
I want to provide you with an update on OpenAFS compatibility and Windows Vista. A couple of months ago a bug was discovered in the Vista NetBIOS stack that would cause OpenAFS to crash anytime the operating system was placed into "Standby" mode. Microsoft took the bug seriously and fixed it. Unfortunately, the produced instability in other areas of the operating system and Microsoft was forced to pull the fix. Therefore, the final Vista release that will be made publicly available in January is going to cause existing OpenAFS versions to crash. As far as I can tell there is no work around to avoid the problem. Microsoft has had an engineer working on the problem for a couple of weeks without success. Microsoft is committed to fixing this bug for Vista SP1. Once a fix is available they will most likely make a QFE available to customers that have support contracts. This has left OpenAFS in an awkward situation until a fix is available. The best that I believe I can do is cause OpenAFS to gracefully shutdown when this error condition is detected. The benefit of a graceful shutdown is that the data stored in the cache will be preserved. However, if there were applications with files open within AFS at the time of Vista entering "Standby" mode, these files would not be available when the Operating System restarts. It is my hope that Microsoft will allow OpenAFS to distribute the QFE (when available) as part of the OpenAFS installers. If so, the majority of the damage can be avoided. The long term solution is clear. The OpenAFS community needs to move away from the SMB Gateway model and replace it with a native Network Provider / File System Filter model. Completing this work in such a way that it is portable across Windows 2000 SP4 through Vista and works on both 32-bit and 64-bit operating systems is going to be time consuming and expensive. I hired a Windows file system and kernel driver developer to review the work that has already been done and help design an architecture that can support all of the required platforms for the next ten years. We believe that this work can be accomplished over a period of ten months provided that the necessary financial resources ($250,000 - $300,000) can be acquired. The long term benefits of getting rid of the SMB gateway are: (1) the use of the Microsoft Loopback Adapter will no longer be required (2) all of the problems associated with the CIFS client timeout issues will be removed. this will not only improve performance but will solve all of the "delayed write" and "network path no longer available" errors. (3) performance gains. There is significant time delays and CPU load added to each transaction as a result of the CIFS client and SMB server both involving the SMB/NetBIOS/TCP/IP stack. If your organization is in a position to assist us in financing this work, please contact me off-list. Jeffrey Altman OpenAFS Gatekeeper/Elder smime.p7s Description: S/MIME Cryptographic Signature
RE: [OpenAFS] AFS and PhotoShop interaction?
What does traceroute or ping show? Here its .28ms on the intranet. 100BT or 1,000BT network? One might also create a metric for "slow" and "fast". tedc -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Jeffrey Altman Sent: Tuesday, November 07, 2006 5:02 AM To: [EMAIL PROTECTED] Cc: Christopher D. Clausen; openafs-info@openafs.org Subject: Re: [OpenAFS] AFS and PhotoShop interaction? Mike Polek wrote: > Christopher D. Clausen wrote: >> Mike Polek <[EMAIL PROTECTED]> wrote: >>> information and would like to know if anyone else has had a similar >>> experience. This is on Windows XP using 1.4.2a-DEBUG client. >>> The fileserver is running Fedora Core 5 and OpenAFS 1.4.2 (new >>> machine). Similar behavior using 1.4.0 and 1.4.1 windows clients and >>> 1.4.1 fileserver. >> >> >> I'd suggest trying the 1.5.10 client. > > I have a couple people with 1.5.10 clients installed, and they > experience the same behavior. I am now borrowing a WinXP box > and it's sitting on my desk. I experience the same behavior. Follow the directions in the release notes for capturing log data with the SysInternals FileMon and DbgView tools. FileMon will show what operations the applications are requesting. DbgView will capture what OpenAFS is doing in response. Set both tools to capture with millisecond clock time and the logs can be examined in lock step. Send a bug report with the logs to [EMAIL PROTECTED] Jeffrey Altman ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
Re: [OpenAFS] AFS and PhotoShop interaction?
Mike Polek wrote: > Christopher D. Clausen wrote: >> Mike Polek <[EMAIL PROTECTED]> wrote: >>> information and would like to know if anyone else has had a similar >>> experience. This is on Windows XP using 1.4.2a-DEBUG client. >>> The fileserver is running Fedora Core 5 and OpenAFS 1.4.2 (new >>> machine). Similar behavior using 1.4.0 and 1.4.1 windows clients and >>> 1.4.1 fileserver. >> >> >> I'd suggest trying the 1.5.10 client. > > I have a couple people with 1.5.10 clients installed, and they > experience the same behavior. I am now borrowing a WinXP box > and it's sitting on my desk. I experience the same behavior. Follow the directions in the release notes for capturing log data with the SysInternals FileMon and DbgView tools. FileMon will show what operations the applications are requesting. DbgView will capture what OpenAFS is doing in response. Set both tools to capture with millisecond clock time and the logs can be examined in lock step. Send a bug report with the logs to [EMAIL PROTECTED] Jeffrey Altman smime.p7s Description: S/MIME Cryptographic Signature
Re: [OpenAFS] LiveCD for openafs disaster recovery scenarios ?
Madhusudan Singh wrote: > Hello > > At least one (and possibly two) hard disks on the machine (has RAID 5 - > rebuild did not help too much) that runs our openafs cell failed today. This > is a Debian machine. Partition /usr is down and attempts to mount it fail > with bad superblock messages. As a result, all network and many other system > commands (awk, etc.) are inaccessible. > > Now, is there a live CD that can be used to run the machine so that the users > can at least copy over their data ? I think a reinstall is indicated. > > The two partitions housing /vicepa and /vicepb are not affected. Is there a > way to retrieve data directly from the hard disks ? > > Thanks. > ___ > OpenAFS-info mailing list > OpenAFS-info@openafs.org > https://lists.openafs.org/mailman/listinfo/openafs-info > Can you reinstall the OS and leave /vicepa and /vicepb alone during the install? Then once you are back in business install openafs , mount /vicpea /vicepb and away you go? We do this often however our partitions are on attached jbods so we swap out servers or rebuild the server from an install server but its basically the same thing. It works as long as the the server hostname remains the same. /sd -- Steve Devine Storage Systems Academic Computing & Network Services Michigan State University 506 Computer Center East Lansing, MI 48824-1042 1-517-432-7327 Baseball is ninety percent mental; the other half is physical. - Yogi Berra ___ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
