On 4 Oct 2005, at 10:46, Philip Craig wrote:
The server has 0 users connected and 1 other server connected. It's
a backup server in case of connectivity loss on the main server.
The crashes happen at random, after say 3-6 hours. When the crashes
happen, no one on the server that crashes has done anything,
because it has no users!
Just because it has no users does not mean it's not receiving
commands from other users :) AWAY is propagated across all servers to
keep info for that user in sync.
On the linked server (not the one that is crashing) which is a
small private server (see link details in config in my initial bug
report) there are about a dozen users connected, coming and going
more or less randomly as is normal with irc. I haven't noticed any
pattern on the linked server before the crashing server crashes.
What is different between my two setups (I run both servers under
rageircd) is that the good one is a single CPU AMD64, and the
crashing one is a dual Intel PIII.
In theory rageircd shouldn't have an issue with dual CPUs. In
practice the story can be quite different!!
What I will say is that I've had a server that's been up and down
like you're describing before. However the server has had a lot of
hardware issues so I presumed it was that and gave up. However
perhaps I was wrong.
You mention heap corruption. Is there a way of building my rageircd
sources with some kind of heap checking/validation tool? valgrind
perhaps?
Yeah, replace -O2 with -O0 ('oh zero') in the top-level Makefile and
read the valgrind Quick Start guide. ;-) I will do the same!
I will look into this when I have the chance, probably Friday
afternoon, unless you can come up with anything in the meantime.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
