Re: Re(5): kernel dump server
Re: Re(5): kernel dump server
- Subject: Re: Re(5): kernel dump server
- From: Derek Kumar <email@hidden>
- Date: Wed, 1 Feb 2006 23:18:38 -0500
... (since the lack of ARP resolution will be addressed in a
future update
This is actually the "biggie" for us, and I suspect other
developers too. If you have a full-blown network with routers then
you can put the dump server on another network and things "just
work". But for those of us with smaller networks (even large-ish
companies with fairly simple networks) this is a killer. It makes
kernel-dump entirely useless, so please can we have it in a
"sooner" update.
I must note that this generalization is something of an exaggeration.
Section 5.2.4.2 of RFC 1812 ("Requirements for IP Version 4 Routers")
appears to require compliant implementations to forward packets to
the appropriate attached interface, regardless of the origin of the
packet. Here's an excerpt describing the forwarding algorithm:
5.2.4.2 Local/Remote Decision
After it has been determined that the IP packet needs to be forwarded
according to the rules specified in Section [5.2.3], the following
algorithm MUST be used to determine if the Immediate Destination is
directly accessible (see [INTERNET:2]).
...
# If no network interface has been selected in the first step, for
each IP address assigned to the router:
1. isolate the network prefix used by the interface.
IMPLEMENTATION
The result of this operation will usually have been
computed and saved during initialization.
2. Isolate the corresponding set of bits from the IP Destination
Address of the packet.
3. Compare the resulting network prefixes. If they are equal to
each other, the packet can be transmitted through the corresponding
network interface.
Empirically, on the (RFC 1918 private) network that I'm composing
this message on, where the router is a garden variety cable modem,
crashdumps work fine when both source and destination are on the same
network. The same is true of the routers on the Apple and MIT
campuses. Additionally, I've helped engineers at external
installations on multiple occasions in the past (Cisco and Oracle
spring to mind) where crashdumps have been similarly configured. I
don't believe it is completely unreasonable for a minimalist
implementation to rely on this behavior.
That being said, it is both inefficient and problematic in certain
scenarios (such as yours) to direct packets exclusively to the
router. Also, if there is no router present (as on a test network),
you may have to use the workaround of setting the crashdump server as
your gateway, as has been noted on this list in the past. I can also
appreciate that there are many instances where it is not feasible to
change router behavior (I believe this was the case at your
installation) and, as I've noted previously, this is something we
intend to address in an update as soon as possible.
Derek
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-kernel mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden