[OpenAFS] Windows client and Linux server issues.

[Mitchell Baker]

On Wed, 2006-09-27 at 11:07 -0400, Jeffrey Altman wrote:
> Mitchell:
> 
> Do you have the same issues with the 1.5.8 Windows client?

Just installed on my test system and no, there is not issue there.

> 
> The error messages you describe are the Windows SMB client timing
> out the connection to the AFS Client Service because the AFS client
> is taking too long to respond.
> 
> 1.5.8 implements a new algorithm to attempt to convince the SMB client
> to be more patient.  There is no way to prevent the SMB client from
> timing out because there is no method to inform the client that its
> request is still in progress.  This is one of the reasons why we need
> to implement a Windows File System Redirector in place of the SMB gateway.
> 
> The Solaris 1.2.13 vs Linux 1.4.1 comparisons and the addition of 'r'
> are certainly interesting data points.  Seeing trace logs from the
> Windows client and File Server logs (at level 125) that match would
> be very useful.

Looking for the info on how to get a trace from the windows side of
things.  Did the fs minidump is that what you are looking for?


> 
> Jeffrey Altman
> 
> 
> 
> Mitchell Baker wrote:
> > I'm having an issue which I need some help with.  Just finished
> > migrating all of our file servers from Solaris 8 running openafs 1.2.13
> > server to RHEL 4 x86_64 running 1.4.1.  I have one test server running
> > 1.4.2fc4. 
> > 
> > Windows client used is using AFS version 1.3.8700 and same thing with
> > 1.4.0203. Windows XP sp2.
> > 
> > We have setup "homework turn in" directories in some of the classes.
> > Students only have "li" access into those directories so they can turn
> > in homework, but can't see other students work after it is turned in
> > etc...
> > 
> > This worked fine from the windows client on the solaris servers.  But
> > now under the linux servers, there are huge delays and 50% of the time
> > access failure with a message "(dir name) is not accessible. The
> > specified network name is no longer available.". 1.4 client will give
> > there error: (dirname) is not accessible. You might not have permission
> > to use this network resource.  Contact the administrator of this server
> > to find out of you have access permissions.  The specified network name
> > is not longer available. The event log under windows shows these
> > Warnings:
> > 
> >  The description for Event ID ( 1005 ) in Source ( AFS Client ) cannot
> > be found. The local computer may not have the necessary registry
> > information or message DLL files to display messages from a remote
> > computer. You may be able to use the /AUXSOURCE= flag to retrieve this
> > description; see Help and Support for details. The following information
> > is part of the event: Pkt straddled session startup, took 152125 ms, ncb
> > length 109.  (Several of these with varying time)
> > 
> > The description for Event ID ( 1009 ) in Source ( AFS Client ) cannot be
> > found. The local computer may not have the necessary registry
> > information or message DLL files to display messages from a remote
> > computer. You may be able to use the /AUXSOURCE= flag to retrieve this
> > description; see Help and Support for details. The following information
> > is part of the event: cm_Analyze: HardDeadTime exceeded..
> > 
> > If I add "r" to the ACL all works great. 
> > 
> > If all the files are owned by the users trying to access the directory
> > there are no issues.  Fewer files in the directory make a difference as
> > well.  Fewer number of files, faster listing, fewer failures. 
> > 
> > Again the Solaris servers running 1.2.13 have no issues as all. 
> > 
> > I think I have covered most of my testing...
> > 
> > Any offers on how to correct this problem?
> > 
> > See-ya
> > Mitch
> > 
> > 
> > 
> >