RE: [reSIProcate] do DnsResult::Lookup() for every TransactionState , Is it necessary ?
Another point is that the DNS caching is resiprocate will make most calls to
lookup be a cache hit rather than a query over the wire.
--Derek
> -----Original Message-----
> From: resiprocate-devel-bounces@xxxxxxxxxxxxxxxxxxx [mailto:resiprocate-
> devel-bounces@xxxxxxxxxxxxxxxxxxx] On Behalf Of Jason Fischl
> Sent: Friday, December 02, 2005 7:03 AM
> To: ???
> Cc: resiprocate-devel
> Subject: Re: [reSIProcate] do DnsResult::Lookup() for every
> TransactionState ,Is it necessary ?
>
> On 12/2/05, 贾军营 <jiajunying@xxxxxxxxxx> wrote:
> > Hi,
> > I find that Resiprocate creates new DnsResult object for every
> TransactionState ,so the new DnsResult object will do lookupSRV() for it's
> TransactionState, I don't think it's necessary. I think that it's very low
> effecitve and the DnsResult object can be reused . Am I right?
> > Please advise.
> >
>
> DnsResult contains dns results that are specific to the transaction.
> Since there may be multiple concurrent transactions, how can we reuse
> the same DnsResult object between different transactions?
>
> It might be worthwhile using a pool of DnsResult objects but I've not
> noticed any indications from profiling the stack that would indicate
> that this is the best place to optimize.