Olav Vahtras | 9 Nov 10:10 2006
Picon
Picon

mac osx arla ticket forwarding problem

Skillnad p -f och -F:

$ which klist
/usr/arla/bin/klist

$ klist -f
Credentials cache: API:Initial default ccache
        Principal: vahtras <at> NADA.KTH.SE

  Issued           Expires        Flags    Principal
Nov  9 09:58:17  Nov  9 19:58:16  FI     krbtgt/NADA.KTH.SE <at> NADA.KTH.SE
Nov  9 09:58:17  Nov  9 19:58:16         afs/pdc.kth.se <at> NADA.KTH.SE

$ telnet -f lise.pdc.kth.se
...
[ Kerberos V5 accepted forwarded credentials ]
...
d14n36$ klist
Credentials cache: FILE:/tmp/krb5cc_21885
        Principal: vahtras <at> NADA.KTH.SE

  Issued           Expires          Principal
Nov  9 10:02:05  Nov  9 19:58:16  krbtgt/NADA.KTH.SE <at> NADA.KTH.SE
Nov  9 10:02:06  Nov  9 19:58:16  afs/pdc.kth.se <at> NADA.KTH.SE
Nov  9 10:02:06  Nov  9 19:58:16  afs <at> NADA.KTH.SE

funkar ok

$ telnet -F lise.pdc.kth.se
...
(Continue reading)

Harald Barth | 9 Nov 11:56 2006
Picon
Picon

Re: mac osx arla ticket forwarding problem


> Skillnad på -f och -F:

arla-drinkers is an english speaking mailing list. However,
this is probably a question for the also english speaking
heimdal-discuss mailing list.

The problem here is that telnet/telnetd in heimdal 0.7.2 has a bug
which in certain combinations prevents ticket forwarding to work
correctly between machines with different endianess. Sometimes this
can be worked around by using addressless tickets, sometimes not. I
don't have a list of combinations that work and which won't, the
matrix would include at least 2 versions, 4 combinations of endianess,
2 for tickets with and without addresses... you see where this is
leading. This bug was fixed on the server (telnetd) you connect to. It
is probably not fixed on the client you use, because the dmg which
contains the fix has not been released "publicly" (or what you want to
call it) yet. Tol was very nice and fixed an arla dmg with the
bugfixed heimdal binaries for me, but the right way would be to
release an heimdal dmg with the bugfixed programs needed (for arla,
you only need kinit and afslog, so telnet is just a bonus).

file:///afs/stacken.kth.se/ftp/pub/random/tol/arla-0.43-haba.dmg
      ftp://ftp.stacken.kth.se/pub/random/tol/arla-0.43-haba.dmg

Testing of the above version has been 'limited' (or how you best want
to call it).

I'm still hoping for heimdal 0.7.3 or 0.8 to appear. It would be very
nice to have a heimdal dmg but I doubt we will see that from the
(Continue reading)

Rasmus Kaj | 10 Nov 00:45 2006
X-Face
Picon
Picon

Simple patch for pts.c


In appl/pts/pts.c, the chown_cmd function, there is a mistake in
argument ordering, so pts chown fail.

There is a patch in /afs/kth.se/home/k/a/kaj/Public/patch-ptc.c or
attached to this message that makes pts chown work.

--

-- 
Rasmus Kaj <rasmus <at> kaj.se> -- http://www.stacken.kth.se/~kaj/
All science is either physics or stamp collecting. -E. Rutherford

Attachment (patch-ptc.c): text/x-patch, 845 bytes
_______________________________________________
Arla-drinkers mailing list
Arla-drinkers <at> stacken.kth.se
https://lists.stacken.kth.se/mailman/listinfo/arla-drinkers
Tomas Olsson | 10 Nov 10:45 2006
Picon
Picon

Re: Simple patch for pts.c

Rasmus Kaj <kaj <at> kth.se> writes:
> In appl/pts/pts.c, the chown_cmd function, there is a mistake in
> argument ordering, so pts chown fail.
> 
> There is a patch in /afs/kth.se/home/k/a/kaj/Public/patch-ptc.c
> 
Thanks. The original mistake was most likely in that this is the only place
where the host argument was added after 'cell', so I'd prefer just
reordering the command's arguments for consistency. Would that cause any
problems (like widely deployed arla-dependent scripts for example)?

It would also make it slightly less different from the OpenAFS version,
which is good.

/t
Rasmus Kaj | 10 Nov 11:41 2006
X-Face
Picon
Picon

Re: Simple patch for pts.c

>>>>> "TO" == Tomas Olsson <tol <at> it.su.se> writes:

 >> There is a patch in /afs/kth.se/home/k/a/kaj/Public/patch-ptc.c

 TO> Thanks. [...] I'd prefer just reordering the command's arguments
 TO> for consistency. Would that cause any problems (like widely
 TO> deployed arla-dependent scripts for example)?

No problems that I know of.  And for any unknown deployed scripts it
would probably be better to have them compatible between arla and
openafs instead.

 TO> It would also make it slightly less different from the OpenAFS
 TO> version, which is good.

Agreed.

--

-- 
Rasmus Kaj <rasmus <at> kaj.se> -- http://www.stacken.kth.se/~kaj/
What is the word where la is the middle, is the beginning, and the end?
Tomas Olsson | 13 Nov 10:33 2006
Picon
Picon

Re: Simple patch for pts.c

Rasmus Kaj <kaj <at> kth.se> writes:
> In appl/pts/pts.c, the chown_cmd function, there is a mistake in
> argument ordering, so pts chown fail.
> 
Fixed.

thanks
        /t
Dr. Rainer Rohmfeld | 14 Nov 17:49 2006
Picon

Problem with file creation on openafs-1.4.1

Hello,

we are using an Ubuntu 6.06 LTS system with openafs-1.4.1.
Arla-0.43 compile, after enabling DHAVE_OPENSSL manually in config.h,
and is starting and running. Reading of files is ok.

If the arla-client is writing data to a file on the file-server, the data 
looks like

 <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ...

What is the problem here?

Validated the same effect with arla-0.42.

--

-- 

Best Regards / Mit freundlichen Grüßen

R. Rohmfeld
Tomas Olsson | 14 Nov 20:52 2006
Picon
Picon

Re: Problem with file creation on openafs-1.4.1

"Dr. Rainer Rohmfeld" <Rainer.Rohmfeld <at> emuge.de> writes:
> we are using an Ubuntu 6.06 LTS system with openafs-1.4.1.
>
The server is openafs 1.4.1 and the client is arla on Ubuntu? What is the
kernel version on the client?

> If the arla-client is writing data to a file on the file-server, the data 
> looks like
> 
>  <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ^ <at> ...
> 
Looks like zeroes

Hm. That's very strange. How do you write data, and does it look the same
if you read it from other hosts?

/t
Harald Barth | 14 Nov 23:01 2006
Picon
Picon

Re: Problem with file creation on openafs-1.4.1


> we are using an Ubuntu 6.06 LTS system with openafs-1.4.1.

I'd start by asking the following questions...

What is LTS?
Kernel version?
Is OpenAFS 1.4.1 the server version?
Is the result the same if you write without auth (as system:anyuser)?
How do you write the data? 
strace on the writing command?
Output of fs arladebug almost-all?
Output of fs nnpfsdebug almost-all?
How does the data look from another client?
Can data be written from another client and read correctly?
Can you run the test suite (*)?

cd tests
WORKDIR=irgendwo-unter-/afs-wo-schreibbar ./run-tests -fast -all

Harald.
Harald Barth | 15 Nov 16:59 2006
Picon
Picon

Re: Antwort: Re: Problem with file creation on openafs-1.4.1


Copy to the list...

From: Rainer.Rohmfeld <at> emuge.de
Subject: Antwort: Re: Problem with file creation on openafs-1.4.1
Date: Wed, 15 Nov 2006 11:38:54 +0100

> Harald Barth <haba <at> pdc.kth.se> schrieb am 14.11.2006 23:01:37:
> 
> > 
> > > we are using an Ubuntu 6.06 LTS system with openafs-1.4.1.
> > 
> > I'd start by asking the following questions...
> > 
> > What is LTS?
> It means Long Time System and is Ubuntu specific.
> > Kernel version?
> Validated on 2.6.15 and 2.6.17
> > Is OpenAFS 1.4.1 the server version?
> Yes
> > Is the result the same if you write without auth (as system:anyuser)?
> Yes
> > How do you write the data? 
> copy a file from the client to AFS, create a file with any editor, echo a 
> string to a file on the afs fileserver
> > strace on the writing command?
> No
> > Output of fs arladebug almost-all?
> > Output of fs nnpfsdebug almost-all?
> Attached LogFile
(Continue reading)


Gmane