BIND 9 crapping out at 510 MB
LConrad at Go2France.com
Fri Oct 20 14:34:50 UTC 2006
On two different machines, 2 or 3 times each, when the named
footprint reaches about 510 MB,
queries fail with SRFVAIL,
rndc flush and rndc stop don't work,
rndc status does work
kill named-pid doesn't work
we must "kill -KILL named-pid"
To try to prevent the problem, we have added
to try to flush records at 12 hours max.
More information about the bind-users