bind 9.2.3rc3 problems

[Ladislav Vobr]

>>in a ISP environment with 30000 customers ..about 15 20 minutes online
>>...the named service stops responding to requests ...restarting named v=
> 
> ia
> 
>>rndc reload DOESNT cure the issue ..i found that the named cache is
>>filling up (as well as i read now thats thers memory leak bug in this
>>release of bind.) anywho..

We had similiar problem, after patching the solaris9 with latest 
patches, it disappeared. We have e280 1xcpu, bind either 9.2.1 or 9.2.3 
I don't remember, now its 9.2.3 and it's fine.

or you might check things below...

did you check you recursive-client queue? What's you req/sec rate? Did 
you enable logging for client category? how big is your cache when it 
stops responding? what's the cpu load when it stops? Do you receive 
weird traffic?

Ladislav