Strange memcached behavior on stress test
sgrimm at facebook.com
Sun Jun 11 16:28:40 UTC 2006
> We restart it daily.
We, on the other hand, never restart it, and we transfer way more than
64GB of data with no signs of instability. I just looked at one of our
memcached instances and it has these stats (commas added for readability):
STAT bytes_read 204,016,350,632
STAT bytes_written 11,728,604,145,226
Here's another one that serves more volatile data (bigger ratio of
writes to reads):
STAT bytes_read 2,606,059,831,972
STAT bytes_written 8,071,661,999,891
This is after about 30 days of uptime. We only restart memcached when we
make configuration or code changes; it is rock solid other than that.
Caveat: We are, as you might expect, running the Facebook code, not the
stock 1.1.12 code. So it's possible some of the changes we made had the
side effect of fixing whatever problem you're running into.
More information about the memcached