Re: [JXTA dev] Failure to resolve output pipe

I have seen this too with JxtaSocket which basically uses a pipe  
under the covers. For one 10 secs sounds too short. The peers on the  
same machine don't really know they are on the same machine. They  
still have to follow the same pipe publish and resolution process  
that every peer has to go through. I would up that to 30 or 60 secs.  
Even then, I have seen some instances (few) where pipe resolution  
still times out. For that I added a retry loop which seems to work. I  
know waiting potentially 2-3 mins for pipe resolution is not a  
preferable thing but its the steps we had to take for reliability.

Ash


On Dec 7, 2005, at 11:25 AM, Alex Iribarren wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi all,
>
> I've been noticing this for a while, and I've finally decided to
> ask. I have several peers running on the same machine and they
> behave just like they're supposed to, except that sometimes they
> can't resolve each other's pipes in time.
>
> I'm using something like:
> OutputPipe outPipe =
> netPeerGroup.getPipeService().createOutputPipe(pipeAdv,
> Params.PIPE_TIMEOUT*1000);
>
> where PIPE_TIMEOUT = 10. This sometimes causes a IOException, which
> if I read the API correctly means the pipe couldn't be resolved,
> meaning that the owner of the pipe advertised in pipeAdv could not
> be found. I remember somebody saying in [email protected] that in their
> tests a message took up to 19 seconds to be sent between different
> peers through the Internet. Surely 10 seconds is enough to resolve a
> pipe within my own machine, is it not?
>
> Is this normal? Does it happen to anyone else?
>
> Regards,
> Alex
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.1 (GNU/Linux)
> Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
>
> iD8DBQFDlxsSos7l8uF4XVERApgjAKCps1JXHgNNJqaDYa1BFenvf3VMAgCfRQaB
> Fs0k9Rx0AFiZjkNtc20WX3E=
> =d/M6
> -----END PGP SIGNATURE-----
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
>
>