On Wed, 2008-09-10 at 18:38 -0700, peter baer wrote:
> This never used to happen on
> earlier versions.

Resource list code in memcache changed slightly with 1.3.4 (some memory
leaks were closed).

> I have attached a test program (main.c) showing how I triggered this
> issue. It has a hard min of 32, and sends a "set" and
> a "get", once every 5 seconds.  This triggers 32 quits every 5 seconds.

My bet is that resource list used for the memcache is reaping 32 unused
resources after that sleep of 5 seconds (i.e. 64 - 32). The ttl most
likely has to be in microseconds here (i.e. 120 * 1,000,000, not just
120). Unfortunately, that argument in memcache code is 32-bit int (where
it should be 64-bit), but you should still fit in 120 million.

-- 
Bojan

Reply via email to