memget errors in bindlog

Jim Reid jim at
Tue Mar 21 08:06:14 UTC 2000

>>>>> "Kevin" == Kevin Darcy <kcd at> writes:

    Kevin> Having said that, however, I agree that crashing and
    Kevin> burning is hardly a reasonable way to deal with memory
    Kevin> exhaustion.

Maybe, but it's hard to devise an acceptable alternative. One factor
is the semantics of Unix VM subsystems: processes don't get smaller
when they give up data space. Another is the cost/overhead of cache
cleaning. This could take a long time and while it is in progress the
name server wouldn't be able to answer queries. That could be just as
disruptive as a dead name server, especially if the "discard data from
the cache because I'm running out of memory" scenario has to happen

Providing enough RAM for the name server is just another example of
capacity planning IMHO. If a process was compute-intensive, you'd make
sure that there was plenty of horsepower in the system's CPU(s). And
nobody would seriously consider running real-time video over a 19.2K
dialup link. Why should providing enough hardware be any different for
a memory-hungry application?

More information about the bind-users mailing list