Is this a good setup - Comments welcome on hub-spoke setup

[Juli Mallett]

On Fri, Mar 25, 2011 at 10:49, Junaid Jeewa <Junaid.Jeewa at craneww.com> wrote:
> My LHR client kept crashing after a small period with an error message:
> (added 512MB more the VM and apparently fixed this issue)
>
> [root at lhr-wp ~]# 1301073407.513597 [/assert] EMERG: void
> PipeProducer::produce(Buffer*): Halting: Assertion (!output_eos_) failed at
> ../../io/pipe/pipe_producer.cc:109 in function void
> PipeProducer::produce(Buffer*).

Others have reported this repeatedly and I hoped some recent changes
would have fixed it, but clearly not.  Like you, they have said that
resource constraints seemed to be related, but I remain baffled as to
how that could be the case.  Was it only the LHR system that crashed
that way?  I wonder if the fact that it's the only 32-bit system is
significant.

Anyway, just thinking aloud.  Thanks for the report, and I think the
setup seems good, but you'll likely see better performance (at a cost
of minimally-greater RAM usage) if you modify the server to use a
separate codec instance for each peer.

Thanks,
Juli.