[26029] in Athena Bugs
Re: acrobat update notification
daemon@ATHENA.MIT.EDU (Angie Kelic)
Wed Sep 1 01:00:22 2004
Date: Wed, 1 Sep 2004 01:00:17 -0400 (EDT)
From: Angie Kelic <sly@mit.edu>
To: Alex T Prengel <alexp@mit.edu>
In-Reply-To: <200409010447.i814lwX4004824@astrophel.mit.edu>
Message-ID: <Pine.LNX.4.58L.0409010049120.6627@nighthawk.mit.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
cc: bugs@mit.edu
Errors-To: bugs-bounces@mit.edu
It looks like I seem to be seeing it most consistently
when I'm opening a pdf with mozilla over the network,
it's certainly possible that it's having trouble reading
from afs at the same time that it's trying to both
download the file from the net and open acroread.
Using ping to my afs fileserver to check, it does appear
that in fact while trying to open a file with acrobat via
mozilla, my ping times to my afs fileserver approach (and
sometimes exceed) 1s for a decent time interval (correlated
with the size of the file I'm attempting to open).
It wouldn't be the first time I've seen horrid asymmetric link
and afs interactions, so it's quite possible that is what's
causing the problem.
Thanks,
Angie
On Wed, 1 Sep 2004, Alex T Prengel wrote:
>
> >Given that acrobat just added a fourth and fifth line to my
> >.sw_messages I don't think the problem is gone.
>
> Well, I'm at a loss. Neither Garry nor I can reproduce the
> problem. I've tried many times, on both Linux and Sun... Garry points
> out that the locker being replicated, the change in the
> message-sending program wouldn't have propagated out of the R/W AFS
> volume until after you began seeing the problem... it's pretty
> unlikely that you were running it from the R/W volume...
>
> I have one hypothesis that might explain what's happening. The message
> sender first tries to read the .sw_messages file, looking for 5.0.9upgrade,
> to decide if it should send the message again (and write 5.0.9upgrade to
> the file). There could be a timing effect too. If your network is flaky
> it might not be able to do the read in time (or at all maybe), thinks you
> didn't get the message and sends it again (also writing to the file again).
> Can you check out this possibility? It's all I can think of...
>
> Alex
>
