So I run several MySQL servers at 5.7 on several FreeBSD machines all at 11.3 (ish). I am noticing that, over a period of a month, swap space grows until the machine either has to be rebooted (rare) or the mysql process needs to be restarted (common).
In researching this issue, I came across an olde article:
https://blog.jcole.us/2010/09/28/mysql-swap-insanity-and-the-numa-architecture/
This appears to be the best candidate for describing the problem, since I am seeing the same effect on both cloud and bare metal servers. When I tried to do
this isn't available in FreeBSD. There is a numactl(1) command where I can set various policies, but it's unclear which to try.
Has anyone run into this issue on FreeBSD before? Is there a way to fix the swap issue by intelligent use of numactl(1) or cpuset(1)? Thanks in advance.
In researching this issue, I came across an olde article:
https://blog.jcole.us/2010/09/28/mysql-swap-insanity-and-the-numa-architecture/
This appears to be the best candidate for describing the problem, since I am seeing the same effect on both cloud and bare metal servers. When I tried to do
Code:
numactl --hardwareHas anyone run into this issue on FreeBSD before? Is there a way to fix the swap issue by intelligent use of numactl(1) or cpuset(1)? Thanks in advance.