For those that do not frequent IRC, our server just underwent really 
nasty meltdown that ended in corrupt filesystem. The guys at humhost are 
working on it now and think that all our data is recoverable and it was 
not caused by a faulty hardware.

The way whole thing developed deffinitely points towards complete memory 
starvation, something used up every bit of unallocated memory not even 
leaving enough to start a SSH session. Normaly first suspect would be 
MySQL server, but all SQL requests at our box go through Apache which 
causes it to blow up on lack of resources for the database. In this case 
Apache was working fine, as was PHP plugin (albeit due to memory 
constraints it was able to run only one, very small script).

That leaves only one valid suspect - wesnothd

My guess is (I might be wrong, once we get back up there are sprobably 
some logs to be checked, they might give other villain) that it is 
leaking memory somewhere. It'd be *very* good to find that leak (if it 
exist) before we put it up again, because otherwise we'll crash again, 
and we'll crash as badly. Checking trunk and 1.5.2 wesnothd code is 
probably the most important task that needs to be done for yesterday.

Again, I might be wrong, but leaking wesnothd is only thing I can think 
of to cause a meltdown with these particular symptoms.

----------------------------------------------------
Nie wiesz jak dojechać docelu.pl ? 
Pokażemy Ci adres, odległość, czas przejazdu 
i najbliższą stację benzynową lub dworzec PKP. 
Kliknij: http://klik.wp.pl/?adr=www.Docelu.pl&sid=432



_______________________________________________
Wesnoth-dev mailing list
[email protected]
https://mail.gna.org/listinfo/wesnoth-dev

Reply via email to