--- snip:  Request made twice politely remains unacknowledged ---
The code was designed so that you could replace one RoutingTable
implementation with another.   Could you please back out your changes to
CPAlgoRoutingTable and move your new implementation to a completely
different class?  That we can easily test one implementation against the
other in an apples-to-apples comparison.
--- snip ---
Are you going to fix this?

Am I really the only one seeing this problem?

--gj



On Tuesday 24 December 2002 11:25, you wrote:

> > On Tue, Dec 24, 2002 at 04:06:16PM +0000, Matthew Toseland wrote:
> > On Tue, Dec 24, 2002 at 11:39:58AM -0500, Gianni Johansson wrote:
> > > On Monday 23 December 2002 19:28, you wrote:
> > > > > On Sun, Dec 22, 2002 at 12:27:12PM -0500, Gianni Johansson wrote:
> > > > > Matthew:
> > > > > I am seeing significantly slower performance since the
> > > > > CPAlgoRoutingTable was changed.
> > > >
> > > > Empirical evidence or even a systematic anecdote would be of
> > > > interest.
> > >
> > > 0) My load average is higher by a factor of 1.5 to 2.0
> > > 1) Page loads in fproxy are excruciatingly slow
> > > 2) Load average increases noticably when downloading freesites with
> > > fproxy. This was not the case with the previous implementation.
> >
> > This suggests that your node is doing more work. What is your local load
> > average like? Are you running a permanent node?
>
> I mean the network load average - see the Network Load infolet.
5500 to 6500 queries an hour, which is pretty much the same as it was before.

>
> > > > > The code was designed so that you could replace one RoutingTable
> > > > > implementation with another.   Could you please back out your
> > > > > changes to CPAlgoRoutingTable and move your new implementation to a
> > > > > completely different class?  That we can easily test one
> > > > > implementation against the other in an apples-to-apples comparison.
> > > >
> > > > Easily test the routing table? How, exactly?
> > >
> > > Easy comparison of the effects of the different routing table
> > > implementations on the node's performance.
> >
> > No, it's NOT easy.
> >
> > > Scipients (oskars/ians?) design allows you to plug in any RoutingTable
> > > implementation when the node is constructed.  When you move your code
> > > into it's own class, we can switch back and forth between rt
> > > implementations at startup time with a cli parameter value.
> > >
> > > If the performance problems go away with the old code, then we know
> > > there is still work to be done on your implementation.  If they don't
> > > then we know your rt changes are not responsible for the problem.
> > >
> > > See freenet.node.Main.main:
> > >
> > >
> > > RoutingTable rt = new CPAlgoRoutingTable(routingStore,
> > >                                                      Node.rtMaxNodes,
> > >                                                      Node.rtMaxRefs,
> > >                                                      Core.randSource);
> > >
> > > // Comments left over from when I replaced scipients routing table impl
> > > -- gj //              RoutingTable rt = new
> > > ProbabilityRoutingTable(routingStore, //                               
> > >                             Node.rtMaxNodes, //                        
> > >                                    Node.rtMaxRefs, //
> > >
> > > > > Thanks.
> > > > >
> > > > > --gj
> >
> > --
> > Matthew Toseland
> > toad at amphibian.dyndns.org
> > amphibian at users.sourceforge.net
> > Freenet/Coldstore open source hacker.
> > Employed full time by Freenet Project Inc. from 11/9/02 to 11/1/03
> > http://freenetproject.org/

----------------------------------------
Content-Type: application/pgp-signature; charset="us-ascii"; 
name="Attachment: 1"
Content-Transfer-Encoding: 7bit
Content-Description: 
----------------------------------------

_______________________________________________
devl mailing list
devl at freenetproject.org
http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to