[OpenAFS] Re: OpenAFS-info digest, Vol 1 #882 - 12 msgs

Chris Snyder csnyder@mvpsoft.com
Wed, 30 Oct 2002 13:55:23 -0500


openafs-info-request@openafs.org wrote:
> Send OpenAFS-info mailing list submissions to
> 	openafs-info@openafs.org
> 
> To subscribe or unsubscribe via the World Wide Web, visit
> 	https://lists.openafs.org/mailman/listinfo/openafs-info
> or, via email, send a message with subject or body 'help' to
> 	openafs-info-request@openafs.org
> 
> You can reach the person managing the list at
> 	openafs-info-admin@openafs.org
> 
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of OpenAFS-info digest..."
> 
> 
> Today's Topics:
> 
>    1. Re: afsd dying on win2k (Rodney M Dyer)
>    2. Re: afsd dying on win2k (Ken Hornstein)
>    3. Re: cache performance (Lester Barrows)
>    4. Re: afsd dying on win2k (Derrick J Brashear)
>    5. Re: cache performance (Nathan Neulinger)
>    6. Re: cache performance (Warren.Yenson@morganstanley.com)
>    7. Re: cache performance (Nathan Neulinger)
>    8. Re: cache performance (Derrick J Brashear)
>    9. Re: 
entication weirdness (Charles Clancy)
>   10. Re: Re: Kerberos V and xscreensaver/xlock (Charles Clancy)
>   11. Re: Authentication weirdness (Tino Schwarze)
>   12. Re: Re: Kerberos V and xscreensaver/xlock (Christian Pfaffel)
> 
> --__--__--
> 
> Message: 1
> Date: Tue, 29 Oct 2002 19:32:28 -0500
> To: Derrick J Brashear <shadow@dementia.org>
> From: Rodney M Dyer <rmdyer@uncc.edu>
> Subject: Re: [OpenAFS] afsd dying on win2k
> Cc: openafs-info@openafs.org
> 
> At 11:57 AM 10/29/2002 -0500, you wrote:
> 
>>On Tue, 29 Oct 2002, Rodney M Dyer wrote:
>>
>>
>>> From the looks of it, I don't think anything is going to be done about 
>>
>>the
>>
>>>problem since no one on the OpenAFS group cares anything about
>>>Windoz...
>>
>>I don't think that's clear, but I can tell you I certainly don't have the
>>time to care.
> 
> 
> Just on the side, my colleagues and I think it's funny that you say 
> this.  Are you paid as an OpenAFS help desk person?  You seem to have 
> enough time to respond to just about everything that hits this list.  Do 
> you ever have time for anything else?  ;)
> 
> I can only hope my sting was "mostly harmless", but it was intended to draw 
> out comments on just what is going on in the group relative to Windows 
> support.  Yes, I am VERY appreciative of the support I'm getting out of 
> this list.  On at least a couple of occasions I've gotten good help.  I'm 
> sorry if I offended anyone.  Believe me, the last time we had to get a very 
> small problem debugged in the Transarc client, it ended up costing us a few 
> thousand dollars to get fixed.
> 
> I'm glad to hear from Mr. Phil Moore at Morgan Stanley.  I'm glad to hear 
> that someone is pony'ing up for support.  But, is the version that Morgan 
> Stanley using available as open source?  Can anyone get a copy of it?  Is 
> it a forked version of OpenAFS?  What is different about it?  How much 
> would it cost us?
> 
> We've been in a real push now for over a year to get a single-sign-on 
> system developed between our Windows/UNIX/Mac machines.  Using Kerberos V 
> as the authentication mechanism and AFS as the filesystem, we've managed to 
> glue everything together as a working unit.  It all works great except now 
> we are having trouble weaning ourselves away from the kaserver.  Seems the 
> Transarc/OpenAFS "klog.exe" can't be forwarded to the "fakeka" 
> daemon.  This wouldn't be a problem except that it is a real annoyance for 
> our users to "kinit" then "aklog" at the command line by hand.  And, we're 
> having problems with "aklog" behind a NAT router for some reason I can't 
> fathom (yes, we've tried addressless tickets).
> 
> BTW, for anyone who cares, if you setup cross-realm authentication for an 
> AD domain to a Kerberos V realm, you may have trouble with AD domain file 
> share access.  This seems to be caused by a bug/feature/design flaw in the 
> Kerberos V replay packet detection.  Microsoft and MIT are currently 
> working the issue out.  We still need AD domain shares because we store 
> files and databases there that AFS cannot support because it doesn't have 
> complete record locking capability.
> 
> Rodney
> 
> 
>> I know a couple of people who probably care, but I'm not
>>going to out them; They're welcome to comment themselves or not, and I
>>have no idea if they can, or have the time, to look into this.
>>
>>I don't suppose anyone has an actual recipe for reproducing this, or is
>>this one of those deals where someone should pray that their network is
>>the same as yours?
>>
>>(Yes, now I'm being sarcastic. How about attaching a hub and a machine
>>with tcpdump next to a dying client and seeing what's going on as close to
>>when it dies as possible?)
>>
>>
>>
>>
>>_______________________________________________
>>OpenAFS-info mailing list
>>OpenAFS-info@openafs.org
>>https://lists.openafs.org/mailman/listinfo/openafs-info
> 
> 
> 
> --__--__--
> 
> Message: 2
> To: openafs-info@openafs.org
> Subject: Re: [OpenAFS] afsd dying on win2k 
> Date: Tue, 29 Oct 2002 20:35:49 -0500
> From: Ken Hornstein <kenh@cmf.nrl.navy.mil>
> 
>>We've been in a real push now for over a year to get a single-sign-on 
>>system developed between our Windows/UNIX/Mac machines.  Using Kerberos V 
>>as the authentication mechanism and AFS as the filesystem, we've managed to 
>>glue everything together as a working unit.  It all works great except now 
>>we are having trouble weaning ourselves away from the kaserver.  Seems the 
>>Transarc/OpenAFS "klog.exe" can't be forwarded to the "fakeka" 
>>daemon.  This wouldn't be a problem except that it is a real annoyance for 
>>our users to "kinit" then "aklog" at the command line by hand.
> 
> 
> Rodney, it seems to me like it would be trivial to have kinit call aklog
> after it's gotten you a TGT.  Didn't you even consider trying that?  And
> have you heard the phrase, "If you're not part of the solution, you're
> part of the problem?"
> 
> 
>>And, we're 
>>having problems with "aklog" behind a NAT router for some reason I can't 
>>fathom (yes, we've tried addressless tickets).
> 
> 
> I suspect the problem is related to the fact that some versions of the
> 524 library wouldn't accept an addressless v5 TGT.
> 
> --Ken
> 
> --__--__--
> 
> Message: 3
> From: Lester Barrows <barrows@email.arc.nasa.gov>
> Organization: Asani Solutions, LLC
> To: openafs-info@openafs.org
> Subject: Re: [OpenAFS] cache performance
> Date: Tue, 29 Oct 2002 20:06:33 -0700
> 
> Whenever a file is accessed on the client, I believe it contacts the cach=
> e=20
> manager to ensure that it hasn't changed. Perhaps the cache manager, rath=
> er=20
> than the file server, would be the most authoritative place to collect th=
> is=20
> information.
> 
> Regards,
> Lester Barrows
> 
> On Tuesday 29 October 2002 03:08 pm, Nathan Neulinger wrote:
> 
>>What sort of additional logging are you looking for in the file server?
>>
>>Also, how do you plan on handling the "if it's already in the cache, th=
> 
> e
> 
>>file server probably won't see a request" issue?
>>
>>Adding more logging is relatively easy to do, just come up with a list.
>>
>>-- Nathan
> 
> 
> 
> --__--__--
> 
> Message: 4
> Date: Tue, 29 Oct 2002 22:19:53 -0500 (EST)
> From: Derrick J Brashear <shadow@dementia.org>
> To: openafs-info@openafs.org
> Subject: Re: [OpenAFS] afsd dying on win2k
> 
> On Tue, 29 Oct 2002, Rodney M Dyer wrote:
> 
> 
>>>I don't think that's clear, but I can tell you I certainly don't have the
>>>time to care.
>>
>>Just on the side, my colleagues and I think it's funny that you say 
>>this.  Are you paid as an OpenAFS help desk person?  
> 
> 
> I have 2 jobs, some percent of one of them is devoted to OpenAFS issues
> and the other is also OpenAFS related.
> 
> 
>>You seem to have 
>>enough time to respond to just about everything that hits this list.  Do 
>>you ever have time for anything else?  ;)
> 
> 
> Some.
> 
> 
>>I can only hope my sting was "mostly harmless", but it was intended to draw 
>>out comments on just what is going on in the group relative to Windows 
>>support.  
> 
> 
> A lot, but not coherently organized. Perhaps that is part of the problem.
> 
> 
>>that someone is pony'ing up for support.  But, is the version that Morgan 
>>Stanley using available as open source?  Can anyone get a copy of it?  Is 
>>it a forked version of OpenAFS?  What is different about it?  How much 
>>would it cost us?
> 
> 
> It's not (apparently) forked OpenAFS, we got patches from Morgan Stanley
> for the real OpenAFS windows client (the incident is still open in the
> openafs-bugs queue)
> 
> 
>>glue everything together as a working unit.  It all works great except now 
>>we are having trouble weaning ourselves away from the kaserver.  Seems the 
>>Transarc/OpenAFS "klog.exe" can't be forwarded to the "fakeka" 
>>daemon.  This wouldn't be a problem except that it is a real annoyance for 
>>our users to "kinit" then "aklog" at the command line by hand.  And, we're 
>>having problems with "aklog" behind a NAT router for some reason I can't 
>>fathom (yes, we've tried addressless tickets).
> 
> 
> I suppose replacing kinit with one that does aklog is right out? We
> (OpenAFS) are going to have to deal with this in the near future to
> support the Kerberos 5 bridge proposal support which is actually expected
> to be useful (though not mandatory) in 1.2.8.
> 
> 
> 
> 
> 
> --__--__--
> 
> Message: 5
> Subject: Re: [OpenAFS] cache performance
> From: Nathan Neulinger <nneul@umr.edu>
> Cc: openafs-info@openafs.org
> Organization: University of Missouri - Rolla
> Date: 29 Oct 2002 21:23:15 -0600
> 
> The cache manager is part of the client. So, yes, it is contacted.
> 
> As long as a callback is still present with the server, there shouldn't
> be any communication with the file server. 
> 
> So, one possible solution would be a cache manager debug set (fs setset)
> that had a very minimal amount of logging generated - to where you could
> reasonably run fstrace regularly on clients. i.e. not a full bore -
> every access, just file opens. 
> 
> -- Nathan
> 
> On Tue, 2002-10-29 at 21:06, Lester Barrows wrote:
> 
>>Whenever a file is accessed on the client, I believe it contacts the cache 
>>manager to ensure that it hasn't changed. Perhaps the cache manager, rather 
>>than the file server, would be the most authoritative place to collect this 
>>information.
>>
>>Regards,
>>Lester Barrows
>>
>>On Tuesday 29 October 2002 03:08 pm, Nathan Neulinger wrote:
>>
>>>What sort of additional logging are you looking for in the file server?
>>>
>>>Also, how do you plan on handling the "if it's already in the cache, the
>>>file server probably won't see a request" issue?
>>>
>>>Adding more logging is relatively easy to do, just come up with a list.
>>>
>>>-- Nathan
>>
>>_______________________________________________
>>OpenAFS-info mailing list
>>OpenAFS-info@openafs.org
>>https://lists.openafs.org/mailman/listinfo/openafs-info
>