On Jun 19, 2008, at 15:29, Chetan Sarva wrote:
I wrote a quick little script[1] to aggregate stats on our cluster
and managed to crash 4 of 11 servers after the first run and another
server (same cluster) after another couple of runs. We're running
memcached 1.2.5 compiled on Ubuntu 6.10 / 2.6.17-10-server kernel.
This happened to me once in my dev environment running 1.2.5 on
Ubuntu 8.04 / 2.6.24-16-generic kernel but I've been unable to
reproduce the problem since. I ran about 50k connections at it
without "luck." I thought it may be related to memory issues so I
tried filling the cache at various levels and still could not crash
it.
Anyone else seen this happen?
It's possible this is the result of a bug that was recently fixed.
I've been having my computer stomp away at doing roughly what you
described and haven't had any problems (yet).
Can you see if you can reproduce it with a more recent build?
http://github.com/dustin/memcached
--
Dustin Sallings