memget errors in bindlog
D. J. Bernstein
75628121832146-bind at sublist.cr.yp.to
Wed Mar 22 01:59:29 UTC 2000
Jim Reid writes:
> What if a (non recursing) server runs out of memory when it loads zones?
I've fixed that problem too. My tinydns server starts up instantly, and
loads records upon demand: http://cr.yp.to/dnscache/ad/server.html
> And how does any system administrator know what is a reasonable
> limit for the name server's cache in their environment?
This is like asking whether your CPU should have a 256KB cache or a
1MB cache. For typical workloads, the larger cache is slightly better,
but the smaller cache works just fine.
Of course, if CPUs used a BIND 8 ``die when the cache is full''
strategy, or a BIND 9 ``revert to completely uncached behavior for new
data when the cache is full'' strategy, then a large cache would be
crucial. But CPUs are built more sensibly than that.
As explained in http://cr.yp.to/dnscache/faq/cachex.html#cachemotion,
the dnscache log contains statistics to let you see how much benefit a
larger cache provides. But the default cache size, 1 megabyte, is
reasonable for most systems.
More information about the bind-users