BIND 8 memory leak symptoms

[Mark.Andrews at nominum.com]

> At 09:04 PM 11/19/00, Mark.Andrews at nominum.com wrote:
> >         Well if you apply the tuning above it should plateau after a 1:15
> >         hours of of operation.   Otherwise there are kinks a 1, 2, 3, 24,
> >         48, 72 hours after startup being typical ttls.  The max_cache_ttl
> >         is 7 days.
> 
> Without delving into the source to better understand the caching functions, 
> however, by forcing an artificially low upper limit to any TTLs, it sounds 
> like we're trading off local resident cache size for interactive 
> response.  By increasing the likelihood for cache misses due to the server 
> flushing 1-hour old RRs, I'd expect we're forcing increased network 
> activity and the frequency of RR lookups.

	Correct.  How much of a trade off depends on the distribution of
	ttls in the answers received.

> Do you have any stats or 
> observational info on what performance shifts occur as the max TTLs are 
> lowered?

	This is really dependent on the usage patterns of the caches
	clients. e.g. web servers vs web browsers have very different
	usage patterns.  The first tend to have random lookups all
	over the place which takes little advantage (apart for tlds)
	of a long lived cache, some short term advantage.  The later
	tend to get a bigger advantage from a long lived cache and also
	have a smaller foot print.

> This would appear to be an issue for very large caching nameservers
> .

	Most definitly.  There is also a distinct advantage in seperating
	authoritative functions from the caching functions.

	Mark

> 
> Thanks,
> 
> 
> Jimmy
> 
--
Mark Andrews, Nominum Inc.
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: Mark.Andrews at nominum.com