Hi Masry,
You might be looking at this the wrong way around. For whatever
reason you machine decided that it was bang out of space and nixed the
memcache process. It may have done it in a way that left the unix
socket around. Each php process would have been hammering on that
socket waiting for a response. As soon as memcached came back, they
all cleared.
Cheers
Dave
On 19/11/2007, at 4:49 PM, Masry Alex wrote:
I installed memcached 1.2.2 and I connect using PHP5 using PECL
extension with a unix socket as it's only one machine.
Quote:
memcached -d -u www -m 1000 -c 10000 -s /server/memcached.sock
my average server load is 0.50%
after I run memcache for a day or two then I get server load to jump
suddenly to 100%
I get that on my syslog
Quote:
Nov 2 10:58:44 server kernel: Out of Memory: Killed process 10337
(memcached).
when I restart memcached server returns to its normal load
in most of my caching I set TTL for the cache vars mostly 1 day
