[Evolution] BUG: frozen while looking up new mail

Tue, 02 Mar 2004 14:09:17 -0500

--=-Nrjqm2gwLKDhqjJbBREO
Content-Type: text/plain
Content-Transfer-Encoding: 7bit

Can you attach to evolution with gdb and see what it is actually doing
in all the threads when this "lock" occurs? Afaik, all the code that
does *anything* over the network is done in another thread (ie. not the
GUI thread).

the output of "thread apply all bt" should be enough.

Jeff

On Tue, 2004-03-02 at 10:53 -0800, Florin Andrei wrote:

> I'm using Evolution-1.4.5-7 on Linux Fedora Core 1 (it's the default Evo
> package provided by Fedora), with Cyrus IMAPd 2.2.3. I'm using the IMAP
> protocol, plain text authentication (Evo and Cyrus are on the same box,
> so it does not matter). Evo is setup to check the Inbox every minute.
> The /home/user/evolution folder is on the same partition (indeed, on the
> same SCSI spindle) as the Cyrus spool, because Cyrus on my system is
> setup to use /home/cyrus as the spool (not the default /var/spool/imap).
> I'm running my own MTA, which is Postfix-2.0.11-5, which delivers mail
> to Cyrus via LMTP over a Unix socket (default configs all over the
> place).
> Evo sends out mail via the local MTA, using the "sendmail" executable.
> 
> Everything's mostly fine, with one exception: Evo gets locked up hard
> every time it checks the Inbox. It's only for a few seconds, and after
> that it resumes normal functioning, but during those moments nothing
> works, i cannot even type into the Composer, or if i copy/paste text
> outside Evo and i hit CTRL-V exactly when it's frozen, it will wait
> until the lookup is over.
> This is not a regular "the system is too loaded to serve new requests"
> type of thing. It is a hard lock, the CPU is mostly idle, the SCSI is
> mostly idle, but Evo is 100% unresponsive for like 5 seconds or so.
> 
> It is very annoying.
> 
> Sure, probably it would help if my Cyrus server was a bit faster, but
> anyway, in my uneducated opinion, Evo should not get totally blocked
> while waiting for an event to complete, especially a network event.
> It looks like the user interface and some network stuff is in the same
> program loop, or something, and one blocks the other.
> If that's true and it's not a screwup of my own design, then it's a bug
> that needs to be fixed.
> 
> -- 
> Florin Andrei
> 
> http://florin.myip.org/
> _______________________________________________
> evolution maillist  -  evolution@lists.ximian.com
> http://lists.ximian.com/mailman/listinfo/evolution

-- 
Jeffrey Stedfast
Evolution Hacker - Ximian, Inc.
fejj@ximian.com  - www.ximian.com

--=-Nrjqm2gwLKDhqjJbBREO
Content-Type: text/html; charset=utf-8
Content-Transfer-Encoding: 7bit

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 TRANSITIONAL//EN">
<HTML>
<HEAD>
 <META HTTP-EQUIV="Content-Type" CONTENT="text/html; CHARSET=UTF-8">
 <META NAME="GENERATOR" CONTENT="GtkHTML/3.1.8">
</HEAD>
<BODY>
Can you attach to evolution with gdb and see what it is actually doing in all the threads when this &quot;lock&quot; occurs? Afaik, all the code that does *anything* over the network is done in another thread (ie. not the GUI thread). 
 
the output of &quot;thread apply all bt&quot; should be enough. 
 
Jeff 
 
On Tue, 2004-03-02 at 10:53 -0800, Florin Andrei wrote: 
 
<BLOCKQUOTE TYPE=CITE>
<PRE>I'm using Evolution-1.4.5-7 on Linux Fedora Core 1 (it's the default Evo
package provided by Fedora), with Cyrus IMAPd 2.2.3. I'm using the IMAP
protocol, plain text authentication (Evo and Cyrus are on the same box,
so it does not matter). Evo is setup to check the Inbox every minute.
The /home/user/evolution folder is on the same partition (indeed, on the
same SCSI spindle) as the Cyrus spool, because Cyrus on my system is
setup to use /home/cyrus as the spool (not the default /var/spool/imap).
I'm running my own MTA, which is Postfix-2.0.11-5, which delivers mail
to Cyrus via LMTP over a Unix socket (default configs all over the
place).
Evo sends out mail via the local MTA, using the &quot;sendmail&quot; executable.

Everything's mostly fine, with one exception: Evo gets locked up hard
every time it checks the Inbox. It's only for a few seconds, and after
that it resumes normal functioning, but during those moments nothing
works, i cannot even type into the Composer, or if i copy/paste text
outside Evo and i hit CTRL-V exactly when it's frozen, it will wait
until the lookup is over.
This is not a regular &quot;the system is too loaded to serve new requests&quot;
type of thing. It is a hard lock, the CPU is mostly idle, the SCSI is
mostly idle, but Evo is 100% unresponsive for like 5 seconds or so.

<FONT COLOR="#000000">It is very annoying.</FONT>

<FONT COLOR="#000000">Sure, probably it would help if my Cyrus server was a bit faster, but</FONT>
<FONT COLOR="#000000">anyway, in my uneducated opinion, Evo should not get totally blocked</FONT>
<FONT COLOR="#000000">while waiting for an event to complete, especially a network event.</FONT>
<FONT COLOR="#000000">It looks like the user interface and some network stuff is in the same</FONT>
<FONT COLOR="#000000">program loop, or something, and one blocks the other.</FONT>
<FONT COLOR="#000000">If that's true and it's not a screwup of my own design, then it's a bug</FONT>
<FONT COLOR="#000000">that needs to be fixed.</FONT>

<FONT COLOR="#000000">-- </FONT>
<FONT COLOR="#000000">Florin Andrei</FONT>

<A HREF="http://florin.myip.org/">http://florin.myip.org/</A>
<FONT COLOR="#000000">_______________________________________________</FONT>
<FONT COLOR="#000000">evolution maillist  -  </FONT><A HREF="mailto:evolution@lists.ximian.com">evolution@lists.ximian.com</A>
<A HREF="http://lists.ximian.com/mailman/listinfo/evolution">http://lists.ximian.com/mailman/listinfo/evolution</A></PRE>
</BLOCKQUOTE>
<TABLE CELLSPACING="0" CELLPADDING="0" WIDTH="100%">
<TR>
<TD>
<PRE>
-- 
Jeffrey Stedfast
Evolution Hacker - Ximian, Inc.
fejj@ximian.com  - www.ximian.com</PRE>
</TD>
</TR>
</TABLE>

</BODY>
</HTML>

--=-Nrjqm2gwLKDhqjJbBREO--