Windows Power Management was Re: [OpenAFS] OpenAFS 1.3.73

[James Burns]

Jeffrey,
Just to clarify, I was being purposefully vague because I was looking
for general information on bug reporting rather than actually trying
to report one. I did figure out some more information on this
so-called hanging, but I'm not sure if it's an AFS bug or a Microsoft
one. I found out that my cell (1.2.11 on Linux) was in some sort of
hung state. Attempts to access it via the Linux client returned
immediate connection timeout errors. In Windows however, attempting to
access the cell hung Explorer, apparently indefinitely. Sometimes
bringing up a My Computer view would hang it, sometimes trying to open
a mapped drive would as well. In one instance stopping the service and
restarting it worked (this was before I understood the issue though,
so I'm not as sure of the details). I didn't know the command line to
do the restart so I ended up killing Explorer a few times before I
realized that I needed to restart my cell (it's a "home" one server
cell). I'll see if I can get my cell hung again (it seems to be
happening when I move large files around) and perhaps send you
connection information so you can see if it happens to you. The
hanging was happening when I was accessing it over the Internet as
well as locally so there's a fair chance you'll see it. I'll send a
bug report after another iteration of trying to get more information.

-James


On Mon, 01 Nov 2004 11:31:57 -0500, Jeffrey Altman <jaltman@columbia.edu> wrote:
> James:
> 
> Bugs are filed by sending e-mail to openafs-bugs@openafs.org.
> The bug report should attempt to include an unambiguous description of
> the problem and how it can be reproduced.
> 
> For example, the term "hanging" is extremely ambiguous.  It does
> not describe the problem.  What does it mean?  What does it refer to?
> 
> Now in the case of a report I am very sure I do know what you are
> referring to.  The purpose of the PM code is to write all dirty buffers
> back to the file servers prior to the machine being suspended or
> shutdown.  Depending on the state of the network this may take time
> or might not be possible at all.
> 
> If your "hanging" means you say you want to suspend and the computer
> waits a minute before doing so then the PM code is currently attempting
> to do something and failing.  If your "hanging" is something else
> you should explain it in a more precise technical description.
> 
> The PM code in the Windows product was implemented at the wrong layer.
> Instead of being added to the SMB client layer it should have been
> added to the cache manager.  The ways things are now the code will
> not work properly on Terminal Server machines nor will it necessarily
> do the right thing in other situations.  In the OpenAFS Request Tracker
> there is an open ticket for this:
> 
>   15523: Windows Power Management fails to do the correct thing
>     http://rt.central.org/rt/Ticket/Display.html?id=15523
> 
> Jeffrey Altman
> 
> James Burns wrote:
> 
> > Jeffrey,
> > Do you have a check list of the steps a person should go through to
> > submit a bug? I'm still seeing issues probably related to power
> > management on my notebook in 1.3.73, though quite significantly less
> > than any previous release. It isn't crashing, but rather hanging, if
> > that makes a difference in the instructions. Thanks for your awesome
> > work on this software.
> >
> > -James
> 
> 
>