[23344] in Athena Bugs
Re: sun4 9.2.13: evolution
daemon@ATHENA.MIT.EDU (Greg Hudson)
Wed Jul 30 01:53:37 2003
Date: Wed, 30 Jul 2003 01:53:34 -0400
Message-Id: <200307300553.h6U5rYMq029375@error-messages.mit.edu>
From: Greg Hudson <ghudson@MIT.EDU>
To: bugs@MIT.EDU
CC: "Jean Foster" <jfoster@MIT.EDU>
In-reply-to: <01b701c35530$87a42aa0$71029812@thira>
After some investigation, here is what I know about this bug:
* As Jean described, the Evolution process will regularly stop
responding to the X server for long periods of time; in fact, I
didn't see it really reach the level of being "responsive,"
although it would draw itself initially and occasionally update
the display.
* She is seeing the problem on an Ultra 5 with 128MB of memory.
However, the machine does not make audible disk noises while
Evolution is unresponsive, so it doesn't seem to be a memory
issue. (Evolution was consuming on the order of 40MB of memory
and not apparently growing; other GNOME processes were about as
big.)
* I attached to an unresponsive process in gdb and a stack trace by
interrupting it. I didn't save the whole trace, but it appeared
to be performing a flush_updates() operation. It did not appear
to be hung in an infinite loop; by repeatedly running "fin" and
waiting, I was able to get the process to update the display again.
* Jean has a large number of IMAP folders (at least 25). I suspect
this is a large part of why the problem seems to be specific to
her so far.
* evolution spews errors like the following (there were 25 of them
in the file of stderr output she sent me):
(evolution-1.4:25669): evolution-shell-WARNING **: e_folder_tree_add() -- Trying to add a subfolder for a path that already exists -- /INBOX/ACST/3rdParty
* I tried updating a cluster Ultra 5 with 128MB to 9.2.14 and
running Evolution and it was perfectly zippy for me. (I did not
try creating and populating a large number of IMAP folders,
though.)
* I have a system call trace from Jean of an Evolution process,
which I have looked through but not drawn any conclusions from
yet. (It's a little hard to tell what's going on because all the
data which gets passed around tends to be X events or CORBA
messages, neither of which are human-readable without tools.)
* Jean claims Evolution 1.4 was more usable for her last week,
before the 9.2.14 update. The only thing which changed in 9.2.14
relevant to Evolution was a libbonobo upgrade (to fix a bug which
prevented the composition of HTML mail); while that could have an
effect, I suspect an increase in mail store volume or some other
factor is the real issue.
* Ricky Rivera, who also uses Evolution on Solaris (which I knew
because of release-77 [3939]), does not see this problem.
At the moment, I believe this problem is unlikely to be widespread and
should not delay bringing the current dev cell improvements to field,
though it does trouble me. I will continue to investigate.
