apollo@ecf.toronto.edu (Vince Pugliese) (12/31/90)
thank in advance. .vp p.s. if there is enough interest i will summarize and repost.
apollo@ecf.toronto.edu (Vince Pugliese) (12/31/90)
over the next month or so we will be connecting, via ethernet, our apollo network (token-ring;dn4500 configured as a gateway) to a couple of sparc machines. realizing that most if not all of the apollo tcp-ip software was(is) busted i obtained a patch tape (october/90) which is supposed to address all of these problems. Question 1: is the above a safe assumption or, is there a later patch tape which fixes a vital piece of tcp-related software? Question 2: are the instructions in the Configuring and ... TCP/IP still valid and correct or is there a new set (e-mailable or ftpable??) that reflects more accurately real-life user experience? thanks in advance, .vp p.s. if there is enough interest i will repost
apollo@ecf.toronto.edu (Vince Pugliese) (12/31/90)
Situation: apollo network running DOMAIN over token-ring dn4500 configured as gateway with ethernet card Question 1: have patch tape of october 1990. is this tape current enough or do more recent tapes fix any of the tcp-ip related stuff? Question 2: is there a list available (ftp or e-mail) of instructions that more properly reflect real-life user experience in setting up tcp-ip? these would supplant those in the apollo docs. thanks in advance .vp p.s. i will repost if there is enough interest
csfst1@unix.cis.pitt.edu (Charles S. Fuller) (01/01/91)
In article <1990Dec31.031121.8773@ecf.utoronto.ca>, apollo@ecf.toronto.edu (Vince Pugliese) writes: > > realizing that most if not all of the apollo tcp-ip software was(is) > busted ... > Question 1: is the above a safe assumption or, is there a later patch tape > which fixes a vital piece of tcp-related software? Not knowing what level of the OS you are running, it's difficult to give a short answer, so ... Under SR9.x, TCP/IP was, indeed, prone to failure for a number of reasons. It did (and still does, at our site) do the job, but resembles nothing that you'll be familiar with in the non-Apollo world. In our current usage of this software, the SR9.x TCP is used to allow E'netted, non-Apollo workstations at a remote site to pass IP packets through an Apollo token ring co-located at that site (running 9.7), across a Domain/Bridge-A link, through our token ring, and onto our E'net. Once configured, it does, in fact, move IP packets. The "SR10.>0" TCP/IP implementation is much more elegant, and actually "looks" like most of the other TCP/IP's out there. There's actually an /etc/hosts, not a /sys/tcp/hostmap/local.txt . And it's fairly robust. You should have no difficulty communicating with SPARCs, or anything else, for that matter. We routinely talk with machines from a number of vendors, running God-knows-what TCP/IP, with no problems. One major "gotcha" with SR10.2 is the "pty corruption" problem. I mention it here because it will affect telnet access to your Apolli. If you're going to be running SR10.2, be prepared for users being unable to login to the machine until you destroy the pty files and re-create them. Frequently. This problem is fixed in 10.3, by the way. > Question 2: are the instructions in the Configuring and ... TCP/IP still > valid and correct or is there a new set (e-mailable or ftpable??) that > reflects more accurately real-life user experience? From my experience, the manuals released with "SR10.>0" are very useful, even to a novice IP administrator like me. Make sure you have the proper manuals for your level of the system, though; the SR9.x manuals are completely useless for SR10 TCP/IP. As for being "e-mailable or ftpable" ... manuals are kinda big; I doubt you'll find them on-line anywhere ... but good luck. Chuck Fuller fuller@nye.nscee.edu
hanche@imf.unit.no (Harald Hanche-Olsen) (01/01/91)
In article <74452@unix.cis.pitt.edu> csfst1@unix.cis.pitt.edu (Charles S. Fuller) writes:
One major "gotcha" with SR10.2 is the "pty corruption" problem. I
mention it here because it will affect telnet access to your Apolli.
If you're going to be running SR10.2, be prepared for users being unable
to login to the machine until you destroy the pty files and re-create
them. Frequently. This problem is fixed in 10.3, by the way.
That is not my experience. We still see it under 10.3, but it seems
to have become more rare. To stop the problem from getting out of
hand we have a cron job removing and rebuilding all ptys once a week
(on Sunday morning).
- Harald Hanche-Olsen <hanche@imf.unit.no>
Division of Mathematical Sciences
The Norwegian Institute of Technology
N-7034 Trondheim, NORWAY
austin@METO.UMD.EDU (Austin L. Conaty) (01/01/91)
I am interested! ----------------------------------------------------------------------- Austin L. Conaty austin@meto.umd.edu Deptartment of Meteorology 301-405-5357 University of Maryland or: HEY YOU!!!!! College Park, Md. 20742-2425 -----------------------------------------------------------------------