Re: [Evolution] EVO worked well today, but ...

[Paul Smith <paul mad-scientist us>]

On Wed, 2008-05-21 at 05:36 -0700, Axel wrote:
> Ok today was the first day for a while, that Evo worked very well in
> the Exchange mode - no issues at all today. Very nice. 
> Yesterday I had some issues with the backend processes disconnecting
> again, funny enough only when I went to the calendar. Now what changed
> until today. Today there was an update, but I did not recognize any
> Evo lib's etc ... 
>
> Now I am wondering whether there is a dependency to a longer delay in
> the exchange server or a specific error message, which could cause
> Evolution to lose contact withe the backend processes. My exchange
> server is on the other end of the world and networks delays do happen
> of course.

Most likely you're seeing this problem:

http://bugzilla.gnome.org/show_bug.cgi?id=512605

I know it doesn't seem the same at first glance but I've been running
evolution-exchange-server under the debugger and this LDAP crasher shows
up in all sorts of places; Evo is doing LDAP lookups for lots of
different reasons.

There's definitely some kind of timing component to it.  I can't find a
reliably reproducible test case, but sometimes it will happen to me 3-4
times a day and other times not at all.  For me, though, the FASTER the
response the more likely to get a crasher.  The slower my machine, or my
network interface (it's much harder to repro from home through my VPN),
or if I enable lots of debugging etc., it happens much less often.

And, I've not seen it happen once since I started running under valgrind
(but that's a heavy price to pay :-)).