[Mono-list] [Fwd: issues mono 2.8(.1)]

[Peter Hagen]

Hi again,

I have some extra information. When one of the requests seems to hang,
which always seems to be a postback, and I have a second tab with the
same website and I just refresh or open again (not necessarily a
postback) the hanging window will suddenly give a "Internal Server
Error" and in the logs I see a:

[Tue Nov 30 12:59:08 2010] [error] (104)Connection reset by peer:
read_data failed
[Tue Nov 30 12:59:08 2010] [error] Command stream corrupted, last
command was -1

It doesn't happen on the same situation. The second tab loads quickly,
so its not that the site is being restarted.

Any ideas?

Cheers

Peter

-------- Forwarded Message --------
From: Peter Hagen <peter at wingsofdeath.nu>
To: mono-list at lists.ximian.com
Subject: issues mono 2.8(.1)
Date: Tue, 30 Nov 2010 12:17:38 +0100

Hi

I have been testing out the 2.8+ release for my Asp.net application,
running with mod-mono. At first the site looked faster then before, but
some times I notice that (at random) a page stops loading, and after a
while (10+ seconds some times) it goes on. Mostly on giving a refresh it
loads again. Could this have anything to do with the stop_world() and
restart_world() principle of the new Garbage Collection?

Yesterday I tried the site with SGEN by modifying
the /usr/bin/mod-mono-server2 like this:

exec /usr/bin/mono --gc=sgen $MONO_OPTIONS
"/usr/lib/mono/2.0/mod-mono-server2.exe" "$@"

and the result was eventually, that my entire memory was been eaten by
the mono process. Is this a known issue? Maybe it was just an issue
because of not completely restarting apache or something. I didnt dare
to try it again, cause its a live site. 

In several different situations (with different sites on different
machines) I noticed the stuttering of the website. This is all with
2.8.1 (2.8.0 also did it) and one machine is a Debian 4 version while
the other is a Ubuntu 10.4 running in VirtualBox. It happens with and
without SGEN enabled. Any ideas about this?

Cheers

Peter