Re: making progress, but Beagle still breaking

[Charlie Law <charlielaw luxsci net>]

Joe, I hope this info is helpful. Thanks for your reply. I really like  
Beagle and am hoping the best for it.

According to System Monitor, the PID for mono-beagled is 3027. Indented  
below that process is another "mono-beagled" listing, with the PID  
3029.  (More PID info is below.)

Yes, from what I can tell my problem looks the same as the one reported  
in 165819. My error messages are nearly identical, and the behavior  
seems to be, too.

FYI, I'm using the Beagle built from CVS on 5 April 2005. The crash  
seems to occur when I'm using Best, but, then, I've not had Beagle  
running without Best, though I can do this if you like. I can't figure  
out what's triggering the crash, however. It does seem to happen when I  
ask Best to show me search results from something that apparently isn't  
in the index yet.

Please let me know how I can help pinpoint this problem further.

I'm running Slackware current, dbus 0.23.4, gtk-sharp-1.0.5,  
mono-1.1.4, gecko-sharp-0.6

FYI, the PIDs for the following processes are:

dbus-daemon-1 = 2990
dbus-launch = 2988
sh (under "mono-beagled 3029" under "mono-beagled 3027") = 3036
mon (under "sh") = 3037.

I looked at /proc/3027/fd and all I see is a bunch of "@1", "@2", etc.,  
listings on up to "@999".



On 04/06/2005 05:39:55 PM, Joe Shaw wrote:
> Hi,
>
> On Wed, 2005-04-06 at 15:39 +0000, Charlie Law wrote:
> > With help from several of you, I've finally got Beagle working
> fairly
> > well on my Slackware 10.1 system, but it still breaks down after a
> > while. Here's the error message that I get:
> >
> > <<
> > Unhandled Exception: System.IO.IOException: Win32 IO returned
> > ERROR_NO_MORE_FILES. Path: /proc/loadavg
>
> Hmm, this could be a dup of bug 165819 and/or 171516.  Can you find
> out
> the pid of the beagled process (the one with a controlling tty, not
> the
> helper process that mono spawns) and look at /proc/<pid>/fd and see if
> you notice a trend?  Check out the attachment in 165819 and see if
> it's
> similar with a bunch of index segment files open but marked as
> "(deleted)".  If this is it, and you can duplicate this reliably
> that's
> great.  We haven't been able to track this bug down for some time.
>
> We're counting on you, Charlie!
>
> Thanks,
> Joe