On 05/07/11 06:25, Bind wrote: > -----Original Message----- > From: "Bind" <b...@dci.ir> > To: "Mark Andrews" <ma...@isc.org> > Date: Tue, 05 Jul 2011 09:55:03 +0430 > Subject: Re: Fwd: Re: Difference between netstat & rndc status > > > Thanks for your best support and answers all the time. > Could u explain more about this list. how it built and when it refreshed? > Regards >
When a client makes a query of a recursive server, sometimes the answer can be given immediately from cache (or from authoritative data if the server is both authoritative and recursive). However, if the server needs to perform iterative resolution of the query on behalf of the client by sending it's own queries to authoritative servers elsewhere,then it has to keep a record of the original client query while this is happening so that it can respond when the answer is available. If the same client makes multiple queries for different names, then these are all recorded separately since each is due its own answer. When the server responds to the client, then the entry is removed from the list. If a client doesn't see the server response, it doesn't mean that the entry isn't removed from the list, it means that the server responded after the client timed out (this sometimes happens if there are problems resolving the query). Duplicate client queries are identified (for example if the client retries) and stored as a single 'recursive client'. If the list reaches the 'soft' limit (this is the number in the middle) then named starts timing out (SERVFAIL) the longest waiting client queries earlier than it would normally in order to make room for new queries. If the list reaches the 'hard' limit, then there is nowhere to put new queries and named will not be able to process them. _______________________________________________ Please visit https://lists.isc.org/mailman/listinfo/bind-users to unsubscribe from this list bind-users mailing list bind-users@lists.isc.org https://lists.isc.org/mailman/listinfo/bind-users