[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ANTS-2.0 DANTE Dies



> >Actually, I don't think this will fix it completely since its only
> >removing the bogus OutChannel from the dante application and not every
> >protocol/application in the system.  So, if it works enough for you
> >now just leave it in there and I'll try to come up with a real fix.
> 
> 	I suspect that the DANTE application is the only one that
> currently remaps virtual IP addresses to different physical IP
> addresses/ports over the lifetime of the ANTS VM.  Is there anything
> else doing something similar?

The regular routing table suffers from the same affliction so it really
does kinda effect everything...

> 	I wonder (not haveling examined the code) whether the
> OutChannel itself should be shut down when it is removed from
> outChannels.  On the one hand, it might be consuming a resource from a
> constrained pool.

Yes, the channel should be destroyed, but the current version of the
jnodeos doesn't make destroy() publicly available and it doesn't have a
finalizer to kill it either (hmm... need to fix that).

>  On the other hand, might there be synchronization
> problems in shutting down the OutChannel?

You're supposed to be able to delete them while the flow is running.  So,
the inevitable synchronization problems will just have to be fixed.

> 					Craig Milo Rogers

tim stack




[ Janos ] [ OSKit ] [ Network Testbed ] [ Flick ] [ Fluke ]
Flux Research Group / Department of Computer Science / University of Utah