memory consumption limit?

[JINMEI Tatuya / 神明達哉]

>>>>> On Fri, 09 Sep 2005 15:47:54 +0200, 
>>>>> Pavel Urban <urbanp at mlp.cz> said:

> we're running Bind 9.3.1 on RedHat Linux version 3, ES. After several 
> days we end up with all memory consumed and DNS nearly unresponsive. We 
> have one master and two slaves; this is a part of our configuration from 
> one slave:

(snip)

> Can you give me a tip what to tweak, or a link to the proper article/web 
> page? Thanks!

The server seems to keep busy for purging some cache entries due to
'over memory' triggered by max-cache-size.

As already discussed in another thread of this list, there are two
things that can generally help in this type of problems:

- use ISC_MEM_USE_INTERNAL_MALLOC
- decrease DNS_CACHE_CLEANERINCREMENT

(see the other thread for more details)

Be careful, however, about the former if you enable threads, since it
will require additional locks which may even decrease the total
performance (and as far as I know this is the reason why
INTERNAL_MALLOC is disabled by default).

Even if these improve your situation, I guess you'll also need to
increase max-cache-size (as long as the physical memory size allows
that), though.  Purging cache entries due to over memory events is
heavy anyway, and it will happen more frequently with smaller
max-cache-size.

					JINMEI, Tatuya
					Communication Platform Lab.
					Corporate R&D Center, Toshiba Corp.
					jinmei at isl.rdc.toshiba.co.jp