I have created

  https://rt.central.org/rt/Ticket/Display.html?id=131726

to track this issue.  There are no obvious changes between 1.7.23 and
1.7.24 which would impact admin vs non-admin access to a file system.
It sounds like a pioctl is failing but not all pioctls are failing which
is very odd.  Attaching a kernel debugger to the system and adding break
points in the pioctl processing pathway will likely be necessary to
identify the issue.

I will attempt to reproduce and address the issue before the 1.7.27 release.

Jeffrey Altman


On 9/10/2013 8:36 AM, Kostas Liakakis wrote:
> Hello again,
> 
> Following up myself, I started trying earlier versions to see if there
> is any change. Mind you, this is the 32bit version installer.
> 
> 1.7.25 Same problem as the latest version
> 
> 1.7.24 Same problem as the latest version
> 
> 1.7.23 Works like a charm
> 
> 1.7.22 BSOD right after logging in. I had the time to briefly see the
> padlock icon with what looked like a black dot at the lower left corner.
> The BSOD message was:
> NO_MORE_IRP_STACK_LOCATIONS
> STOP 0x35 (0x85952130, 0, 0, 0)
> 
> 1.7.21 Works like a charm
> 
> 1.7.20 Works like a charm
> 
> Now I am off to the changelogs to see if I can nail something useful.
> 
> -Kostas
> 
> 
> On 09/09/2013 10:35 PM, Kostas Liakakis wrote:
>>
>> Hello,
>>
>> It has been many years now since we first installed OpenAFS + KfW on our
>> Windows XP clients and have been using it without any trouble. Sticking
>> to the "if it ain't broken, don't fix it" principal we sticked with some
>> 1.3.7x version of the OpenAFS client.
>>
>> Today, as part of a general update, I decided it was time to move on,
>> thinking the IFS redirector had more than enough time to mature. I went
>> on installing Heimdal 1.5.1 binaries, along with NetIDMgr 2.0 and
>> OpenAFS 1.7.26 (in that order) on an otherwise freshly installed WinXP
>> computer which basically only had SP3 and whatever dependencies the
>> CM2012 client brought along (Silverlight etc). Some domain-wide group
>> policy has also been applied, but as this machine belongs to a TEST OU
>> for the time being, they should be mostly harmless stuff.
>>
>> I quickly found about the missing 5-to-4 .DLLs NetIDMgr would require
>> that was not present in Heimdal, complemented them from KfW3.2.2 and
>> soon I was able to get tokens for my realm and browse the AFS
>> filesystem....
>>
>> ... with an administrator account. As soon as I logged on as a simple
>> user, I could not get tokens anymore and the AFS padlock icon would
>> appear broken. While hovering over, a tooltip saying "OpenAFS service
>> cannot be reached" would come up.
>>
>> aklog from command prompt after grabbing tickets from my KDC would say:
>>
>> C:\Documents and Settings\kostas>aklog -d
>> Authenticating to cell physics.auth.gr.
>> Getting v5 tickets: afs/[email protected]
>> Getting v5 tickets: [email protected]
>> pioctl Redirector is ready
>> pioctl NetbiosName = "AFS"
>> pioctl filename = "\\AFS\all\_._AFS_IOCTL_._"
>> About to resolve name [email protected] to id
>> Id 10002
>> Set username to [email protected]
>> Setting tokens.
>> pioctl Redirector is ready
>> pioctl NetbiosName = "AFS"
>> pioctl filename = "\\AFS\all\_._AFS_IOCTL_._"
>> aklog: Cache Manager is not initialized / afsd is not running while
>> setting token for cell physics.auth.gr
>>
>> However, the afsd_service.exe process was running fine.
>>
>> Without logging off, I opened a new CMD window with administrator
>> privileges and was able to get tokens alright and browse the filesystem
>> from the privileged process, so the AFS is working alright.
>>
>> But at the same time, as the non-privileged user, I can't browse
>> \\AFS\all, nor \\AFS\physics.auth.gr which is available even to
>> non-logged users. When I try to do so, Windows responds with:
>>
>> "\\afs\all is not accessible. You might not have permission to use this
>> network resource. Contact your ... "
>>
>> However, I CAN browse \\afs and see the freelance generated root.cell
>>
>> It fells like this is a permission thingie, but I don't really know
>> where to look at right now. Any insight you may have is much appreciated.
>>
>> Thanks,
>>
>> -Kostas
>>
>>
>>
>>
>>
>>
>>
> 

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

Reply via email to