[303] in Zephyr Mailing List
Re: undead zwgcs
daemon@ATHENA.MIT.EDU (Steven Freed)
Thu Sep 17 09:02:56 1998
Date: Thu, 17 Sep 1998 07:00:00 -0600 (MDT)
From: Steven Freed <sfreed@gilasoft.com>
Reply-To: sfreed@gilasoft.com
To: Greg Hudson <ghudson@MIT.EDU>
Cc: zephyr@MIT.EDU
In-Reply-To: <Pine.BSI.4.02.9809161332430.28259-100000@taos.gilasoft.com>
On Wed, 16 Sep 1998, Steven Freed wrote:
> It looks like that is I put the server name in the zephyr server file 3
> times (the server has 3 interfaces) it will work right in most cases since
> the above code should grab all 3 addrs as it loops through, as long as
> nothing else does a dns querey while zephyrd is starting up. I am a bit
> reluctant to hack on this code since I don't know hwat side effects it
> will have.
Well, it looks like this doesn't work either. Remote zhm's still can't
seem to connect to it. I turned on full debugging and it looks like it
really would want 3 zephyrd's running; one for each interface:
Sep 16 18:59:35 spirit zephyrd[6365]: bdump_offer
Sep 16 18:59:35 spirit zephyrd[6365]: bdump_offer: address is 204.134.12.100/1023
Sep 16 18:59:35 spirit zephyrd[6365]: bdump_get: bdump v1.2 avail 192.160.121.100
Sep 16 18:59:35 spirit zephyrd[6365]: bdump_get: connected
Sep 16 18:59:35 spirit zephyrd[6365]: bdump recv loop
So we are back to where we started with how to do this correctly.
Anyone have a good way to fix this? I know the "hp->h_addr" at around line
1000 in server.c needs to be changed to "hp->h_addr_list", but what else
would need to be done, I'm not sure. I am not that versed in network
coding, so if anyone can lend assistance in making this work, I'd
appreciate it.
The current source tree from MIT seems to compile and work well other than
for this multihomed problem. I just grabbed the source from the AFS
server, and the et and ss sources and did "configure", "make", and "make
install". I think I made 2 minor code chages and that was it. ...to
compile in full debugging took a bit more work, I don't think anyone at
MIT has done that in a while and some of the code was a bit broken. This
was all done on BSD/OS 3.1 and 4.0.
I don't want to put together a release untill the multihomed problem is
fixed. I mean, quite frankly, the code is badly broken in this respect and
needs to be dealt with.
--
Steven.
