rsalz@bbn.com (Rich Salz) (05/18/89)
In <285@xdos.UUCP> doug@xdos.UUCP (Doug Merritt) writes: >Since all of your options are unsatisfactory to many people, you can >assume that ".UUCP" is never going to go away, at least not via those >options. Sure you can; you make it VERY WORTHWHILE for folks to join the standards... Concrete example: It would be trivial for me to set things up so that mail to the moderator of comp.sources.unix will NEVER get a reply unless a valid domain address is sent. I have wanted to take lines like this out of my sendmail.cf for quite some time: R$+<@$+.$+.uucp> $1<@$2.$3> u@aaa.bbb.uucp -> u@aaa.bbb R$-!$+ $@$>6$2<@$1.UUCP> resolve uucp names R$+@$-.UUCP $2!$1 u@h.UUCP => h!u I haven't done it yet, but I might. "If you wanna talk, speak my language." Does that make me an asshole? Probably. Do I care? So far, yes, that's I haven't done it. /rich $alz -- Please send comp.sources.unix-related mail to rsalz@uunet.uu.net. Use a domain-based address or give alternate paths, or you may lose out.
doug@xdos.UUCP (Doug Merritt) (05/19/89)
In article <1728@fig.bbn.com> rsalz@bbn.com (Rich Salz) writes: >In <285@xdos.UUCP> doug@xdos.UUCP (Doug Merritt) writes: >>Since all of your options are unsatisfactory to many people, you can >>assume that ".UUCP" is never going to go away, at least not via those >>options. >Sure you can; you make it VERY WORTHWHILE for folks to join the standards... Please don't misunderstand me, I very much like the idea of domain naming becoming a *universal* standard. I'm just concerned about the mechanics of the process. >I haven't done it yet, but I might. "If you wanna talk, speak my language." >Does that make me an asshole? Probably. Do I care? So far, yes, that's >I haven't done it. I appreciate that you care, and I don't think that your concerns make you an asshole at all; it's an understandable temptation. But look at the situation for poorly connected leaves (the situation I've almost always been in myself). It's usually not easy at small companies to scrape up the disk/modem/phone/manager-approval resources in the first place. Once you do (for it's certainly worth the effort to keep trying) you start asking around for a net link, and often are turned down because "we're too small to support an outgoing feed" or because "we're too large to support a leaf". You finally find a cooperative link to the net, get the software set up, hurray, etc. You try hard to make your software follow the standards, to have your local users be informed about netiquette, etc. In general, you follow all the rules to the best of your ability. And then you find out that you're blowing it because you don't have a domain address. Oops! What did I do wrong? Well, let's see, I can't afford to talk to uunet, so that's out. (Ignorant thinking follows, not having seen recent discussion in this group:) Gee, there's thousands of other .UUCP folks out there too, but I want to do things "right". Yet I can't afford to either link directly to Internet nor to call uunet. Now what? Ok, now I've been informed that all I need is to dig up yet another cooperative site (on the arpanet) to do mail forwarding for me. Sounds easy, right? Well, judging by the number of "no's" I've gotten in past years for a simple local uucp news feed (and sometimes even mail links!), that's not necessarily going to be easy. I guess I'll start trying to figure out who nearby might be willing to cooperate, but can you see from the above why it is that there are so many .UUCP leaves without a domain address??? My naive feeling about this is that a small technical solution would solve it: what if there were a way to get a proper domain address *without* requiring mail forwarding? Technically it seems like a reasonable solution. All you need is for smart mailers (particularly at arpa gateways) to understand how to send to e.g. doug@xdos.COM, right? And that general sort of capability is already widespread. Naturally I would have to go through some sort of procedure to get registered, and if it costs a bit (e.g. that famous $35), no problem. If it were that easy, then you could quite quickly get 95% of the .UUCP sites to become part of the .COM domain. So why not? What am I missing??? Thanks, Doug -- Doug Merritt {pyramid,apple}!xdos!doug Member, Crusaders for a Better Tomorrow Professional Wildeyed Visionary
gore@eecs.nwu.edu (Jacob Gore) (05/20/89)
/ news.sysadmin / doug@xdos.UUCP (Doug Merritt) / May 18, 1989 / >Ok, now I've been informed that all I need is to dig up yet another >cooperative site (on the arpanet) to do mail forwarding for me. [...]that's >not necessarily going to be easy. [...] >My naive feeling about this is that a small technical solution >would solve it: what if there were a way to get a proper domain address >*without* requiring mail forwarding? Technically it seems like a reasonable >solution. All you need is for smart mailers (particularly at arpa gateways) >to understand how to send to e.g. doug@xdos.COM, right? And that general sort >of capability is already widespread. But that is EXACTLY what Internet forwarders are! Registering your domain simply means telling Internet mailers which "arpa gateways", as you call them, "understand how to send to, e.g. xdos.com." You do NOT have to be directly connected to your forwarders! Of course, you'll get your mail faster if you are, but what else is new? Jacob Gore Gore@EECS.NWU.Edu Northwestern Univ., EECS Dept. {oddjob,chinet,att}!nucsrl!gore
doug@xdos.UUCP (Doug Merritt) (05/21/89)
In article <3790007@eecs.nwu.edu> gore@eecs.nwu.edu (Jacob Gore) writes: >But that is EXACTLY what Internet forwarders are! Registering your domain >simply means telling Internet mailers which "arpa gateways", as you call >them, "understand how to send to, e.g. xdos.com." Ok. But I'm still missing one central point: why can't I simply register my machine as "xdos.com", and automatically have Internet forwarders know who I am? Why is it necessary to make a personal arrangement with some such machine? I don't see why it can't be as easy as it is with registering uucp map connections. Is it just due to historical problems that grew out of the old arpanet, that no one has gotten around to changing? Or is there some inherent reason for it? Internet folks seem to universally agree that domains are the way to go. Well, if I could get a real domain *purely* by following a registration procedure, with no "would *someone* please do me a a big personal favor and forward mail" arrangements necessary, then it would seem that most .UUCP sites would be .COM sites very rapidly. What, if anything, is wrong with this picture? Thanks, Doug -- Doug Merritt {pyramid,apple}!xdos!doug Member, Crusaders for a Better Tomorrow Professional Wildeyed Visionary
mcb@ncis.tis.llnl.gov (Michael C. Berch) (05/21/89)
In article <320@xdos.UUCP> doug@xdos.UUCP (Doug Merritt) writes: > [...] > Ok. But I'm still missing one central point: why can't I simply register > my machine as "xdos.com", and automatically have Internet forwarders > know who I am? Why is it necessary to make a personal arrangement with > some such machine? I don't see why it can't be as easy as it is with > registering uucp map connections. Is it just due to historical problems > that grew out of the old arpanet, that no one has gotten around to > changing? Or is there some inherent reason for it? The latter. There's no such thing as "Internet forwarders" in general; what matters is *your* site's Internet forwarder(s), which means the machine or machines that are named in MX records in the name server for your domain. Here's how it works: the servers for the top level domain COM are the root servers of the Domain Name System (DNS). They contain a name server (NS) record that will show the name of the name server for your domain. Let's say that machine is NS.FOO.GOV. A resolver will then contact NS.FOO.GOV looking for an MX (mail exchanger) record for the particular machine in your domain (or a wild card that matches the whole domain). The MX record or records (there may be several) will point at hosts that are forwarders for your machine. This latter relationship is something *you* must arrange with the forwarding site, because all that site's machine is going to get when it's forwarding for you are messages addressed to "doug@xdos.com" suddenly showing up in its incoming SMTP connections. It's gotta know what to do with those pieces of mail, like (for example) sending them direct by UUCP or over a private line, or whatever. So the forwarding site has to specially configure its mailer to handle the messages that it is forwarding from the Internet. There's no way to trivially automate this; it requires that the forwarder has some existing method of sending mail to you (or sets one up at your request), and configures its mailer to special-case your site. The example you give comparing this to UUCP map registration is a reasonable analogy, but remember that before you registered the map entry you had to personally set up all the UUCP connections that are mentioned in it. The same is true for domain registration; you have to set up the forwarding relationship yourself, but once you do so, registering the domain and getting it into the Domain Name System is simple & cheap. -- Michael C. Berch mcb@tis.llnl.gov / uunet!tis.llnl.gov!mcb
henry@utzoo.uucp (Henry Spencer) (05/21/89)
In article <320@xdos.UUCP> doug@xdos.UUCP (Doug Merritt) writes: >Ok. But I'm still missing one central point: why can't I simply register >my machine as "xdos.com", and automatically have Internet forwarders >know who I am? ... How are they supposed to know who you are? (The Internet generally doesn't deign to recognize the existence of the UUCP maps, and even if they did, it is not necessarily true that "xdos.UUCP" and "xdos.com" are the same system.) This is like asking the Post Office to deliver mail to you without telling them where you live. >... if I could get a real domain *purely* by following a registration >procedure, with no "would *someone* please do me a a big personal favor >and forward mail" arrangements necessary, then it would seem that most .UUCP >sites would be .COM sites very rapidly... What good would it do to have a domain if nobody could reach you by using it? Arrangements for forwarding *have* to be made if your domain is to be of any use. If you're asking why nobody is willing to automatically forward mail to any site in the uucp maps, who is going to be masochistic enough to volunteer his machine for such a job? Actually, as I understand it, UUNET is quite happy to handle registration and provide forwarding, but quite sensibly insists that you pay for the service. -- Van Allen, adj: pertaining to | Henry Spencer at U of Toronto Zoology deadly hazards to spaceflight. | uunet!attcan!utzoo!henry henry@zoo.toronto.edu
peter@ficc.uu.net (Peter da Silva) (05/22/89)
In article <1989May21.061438.14394@utzoo.uucp>, henry@utzoo.uucp (Henry Spencer) writes: > How are they supposed to know who you are? (The Internet generally doesn't > deign to recognize the existence of the UUCP maps, and even if they did, > it is not necessarily true that "xdos.UUCP" and "xdos.com" are the same > system.) This is like asking the Post Office to deliver mail to you without > telling them where you live. All it would take is one site to set itself up as the MX forwarder for the *.UUCP domain to make everyone in uucp-land happy, right? And if .UUCP is such a small world compared to the internet that'd be no great load, right? And, wow, uunet already does this for you... and gets paid for the privilege. I don't see the problem. -- Peter da Silva, Xenix Support, Ferranti International Controls Corporation. Business: uunet.uu.net!ficc!peter, peter@ficc.uu.net, +1 713 274 5180. Personal: ...!texbell!sugar!peter, peter@sugar.hackercorp.com.
pleasant@porthos.rutgers.edu (Mel Pleasant) (05/22/89)
The last time there was any discussion about a real .UUCP domain, it was soundly trounced by those in control of the root nameservers. So, it isn't so easy after all!! -- Mel Pleasant {backbone}!rutgers!pleasant pleasant@rutgers.edu mpleasant@zodiac.bitnet
matt@oddjob.uchicago.edu (Matt Crawford) (05/23/89)
) You do NOT have to be directly connected to your forwarders! Of course, ) you'll get your mail faster if you are, but what else is new? There are certain things that *can* go wrong if you aren't. And whatever *can* go wrong ... ________________________________________________________ Matt Crawford matt@oddjob.uchicago.edu
matt@oddjob.uchicago.edu (Matt Crawford) (05/23/89)
In article <320@xdos.UUCP>, doug@xdos (Doug Merritt) writes:
) Ok. But I'm still missing one central point: why can't I simply register
) my machine as "xdos.com", and automatically have Internet forwarders
) know who I am?
Define "who I am". Names are entered in the domain system database as
the keys to various types of resource records (RRs). What kind of RR
should be entered for xdos.com?
The first thing that has to happen is that the keepers of the "com"
domain enter one or more NS (name server) RRs for xdos.com. These
direct queries to whatever host keeps the rest of the information about
xdos.com (and *.xdos.com, if any such names exist). Right away you need
the cooperation of SOMEBODY on the internet, because the root servers
who keep "com" do not want to keep all the data for all the subdomains
also.
OK, so you find somebody who will insert your data in their files and
the root servers name that site in the NS record. Now what? The NS
records show that xdos.com exists, but you want to receive mail.
A randomly selected site on the internet delivers mail this way: first
it looks for an MX (mail exchanger) RR for the given name. If one
exists, it names the INTERNET site to or through which mail should be
forwarded. If there is no MX it looks for an A (address) record for the
target. "Address" means numeric internet address. You don't have one.
Unless you find some machine that will accept mail for "NAME@xdos.com" and
send it by UUCP to "xdos!NAME", there is no way for an arbitrary sending
site to bridge the gap between the internet and UUCP. Much of the
internet has no interest in keeping UUCP map files. (For that matter, a
lot of the UUCP world has no interest in keeping them either.)
) Why is it necessary to make a personal arrangement with some such
) machine?
It isn't. You can make a business arrangement instead. For $35 plus
connect time charges, UUNET will take care of you.
) Well, if I could get a real domain *purely* by following a registration
) procedure, with no "would *someone* please do me a a big personal favor
) and forward mail" arrangements necessary, then it would seem that most .UUCP
) sites would be .COM sites very rapidly. What, if anything, is wrong with
) this picture?
Who's going to forward your mail into UUCP-land? Why should they
bother? Do you want someone to force them to do it for you? What are
you doing for them in return?
________________________________________________________
Matt Crawford matt@oddjob.uchicago.edu
lear@NET.BIO.NET (Eliot Lear) (05/23/89)
Matt's message probably demands a bit of elaboration. It isn't hard to imagine a loop occurring if you allow non Internet sites to forward. Moreover, if you use the current version of sendmail, you might get back a wonderful host unknown error message if you try to mail to contact a forwarder through its MX, when its forwarder is down. {This was a real problem when NORDUNET came on line.} For example: I would mail to user@kre.nl. You would see something like the following for that host. kre.nl IN MX hp4nl.nluug.nl 110 and for hp4nl you might see something like the following: hp4nl.nluug.nl IN A 192.16.184.9 hp4nl.nluug.nl IN MX UUNET.UU.NET 100 If hp4nl.nluug.nl was out, sendmail would return a host unknown. Obviously, this is a bug in sendmail, but one that should be fixed if people are going to ``chain''.
bbh@whizz.uucp (Bud Hovell) (05/25/89)
In article <May.22.01.46.36.1989.17215@porthos.rutgers.edu>, pleasant@porthos.rutgers.edu (Mel Pleasant) writes: > The last time there was any discussion about a real .UUCP domain, it > was soundly trounced by those in control of the root nameservers. So, > it isn't so easy after all!! Mel, I'm frankly a little confused (by my own ignorance, I'm sure), but is the problem of setting up a .uucp domain primarily technical, or - as I tend to infer from your note - primarily political? Bud Hovell :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: : UUCP: {sun!nosun | tektronix!{percival|bucket} | attmail}!whizz!bbh : : TELEX: 152258436 (Whizz/Bud Hovell) VOICE: +1 503-636-3000 : : USPO: McCormick & Hovell, Inc., PO Box 1812, Lake Oswego, OR USA 97035 : :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: "May the source be with you!"
jbuck@epimass.EPI.COM (Joe Buck) (05/29/89)
In article <622@whizz.uucp> bbh@whizz.uucp (Bud Hovell) writes: >Mel, I'm frankly a little confused (by my own ignorance, I'm sure), but is >the problem of setting up a .uucp domain primarily technical, or - as I tend >to infer from your note - primarily political? I'm not Mel, but I'll give it a shot. There are both technical and political problems -- the political problems are more of an economic nature, as in Who Pays The Bills.. Yes, the powers that be COULD declare .uucp an official domain, but that would effectively halt the momentum towards the domain conversion of UUCP sites and greatly increase the load on official gateways -- would we have MX records for *.uucp pointing to uunet exclusively, generating large bills for uunet's paying customers to send mail to folks two hops away? Or twist the arms of folks at six or seven sites and try to get some load balancing? Instead, the requirement of a forwarder for each domain in the "UUCP Zone" tends to ensure that no one is being forced to pay for traffic they haven't agreed to. The technical problem is that the "UUCP domain" namespace is flat -- one list must be kept with all the thousands of names, and it must be kept free of duplicates. The list is large enough that errors sometimes creep in (which is why agressive rerouters are so evil, Mel :-). On the other hand, I can name a new PC or workstation here sierra.epi.com or toy-pc.epi.com without having some remote bureaucracy approve, because I'm the domain administrator. -- -- Joe Buck jbuck@epimass.epi.com, uunet!epimass.epi.com!jbuck
roy@phri.UUCP (Roy Smith) (05/29/89)
In <3243@epimass.EPI.COM> jbuck@epimass.EPI.COM (Joe Buck) writes: > The technical problem is that the "UUCP domain" namespace is flat -- one > list must be kept with all the thousands of names, and it must be kept > free of duplicates. [...] On the other hand, I can name a new PC or > workstation here sierra.epi.com or toy-pc.epi.com without having some > remote bureaucracy approve, because I'm the domain administrator. Lest anybody think this is not a real problem (or, to put it another way, not a real advantage of the domain naming system), let me give you an example. We've got a machine here, wombat.phri.nyu.edu. A couple of weeks ago, while I was giving somebody the Grand Tour, I pointed to a rack of stuff in the corner and said "that's wombat, one of our file servers". He replied, "Oh, so *that's* where wombat is! I've sent a lot of mail through there, but never really knew where it was". This seemed odd, since nobody gets their mail on wombat, and it doesn't talk uucp, so nobody would be sending mail through it. It took me a while to realize that he didn't mean my wombat, but some other wombat. Presumably, there is one and only wombat.UUCP, and presumably that was the machine he was thinking of. Wombat being such a nice name, I'm sure there are several wombats in different domains scattered around the world. They all happily coexist, not knowing the they have namesakes in other domains, or caring. -- Roy Smith, System Administrator Public Health Research Institute {allegra,philabs,cmcl2,rutgers,hombre}!phri!roy -or- roy@phri.nyu.edu "The connector is the network"
karl@triceratops.cis.ohio-state.edu (Karl Kleinpaste) (05/30/89)
bbh@whizz.uucp (Bud Hovell) writes:
Mel, I'm frankly a little confused (by my own ignorance, I'm sure), but is
the problem of setting up a .uucp domain primarily technical, or - as I tend
to infer from your note - primarily political?
I wouldn't describe it as a political problem, but realize that
top-level domains like .uucp (not to mention .bitnet, .csnet, .fido,
and .IDontKnowWhatAllElse) all constitute TRANSPORT-based domains,
which is the philosophical inverse of one of the things which domains
are supposed to do: hide the underlying technical nonsense of how the
bits get flown between Hither and Yon.
The folks running the root servers were no more impressed with the
idea of a .bitnet top-level domain than .uucp, for the same reason.
Domains are supposed to provide organizational information, not
connectivity. It shouldn't matter to you what network the remote end
connects with, or how your mail gets there, just so long as it does.
--Karl
Makey@LOGICON.ARPA (Jeff Makey) (06/05/89)
In article <May.22.01.46.36.1989.17215@porthos.rutgers.edu> pleasant@porthos.rutgers.edu (Mel Pleasant) writes: >The last time there was any discussion about a real .UUCP domain, it >was soundly trounced by those in control of the root nameservers. Could "those in control" stop the creation of a UUCP domain if it were called .UU.NET (or something else that fits in the current top level)? I realize that Rick Adams, et al, at uunet.uu.net would have to consent since they "own" that domain name, but would the NIC attempt to stop it? Similarly, there could be .BIT.NET for all of the BITNET hosts. :: Jeff Makey Department of Tautological Pleonasms and Superfluous Redundancies Department Disclaimer: Logicon doesn't even know we're running news. Internet: Makey@LOGICON.ARPA UUCP: {nosc,ucsd}!logicon.arpa!Makey
rick@uunet.UU.NET (Rick Adams) (06/05/89)
The NIC would be happy to register uucp.org (or whatever) for the uucp project. The trick is finding someone stupid enough to be willing to be the uucp forwarder for 15,000 uucp hosts. The fundamental problem is not registering a domain name, but setting up the appropriate forwarders. --rick
tneff@bfmny0.UUCP (Tom Neff) (06/05/89)
In article <56904@uunet.UU.NET> rick@uunet.UU.NET (Rick Adams) writes: >The trick is finding someone stupid enough to be willing to be the >uucp forwarder for 15,000 uucp hosts. > >The fundamental problem is not registering a domain name, but setting >up the appropriate forwarders. Is there no way under the current/proposed scheme to share the forwarding work load in an organized way? The problem as I see it from reading the RFC's and following the discussion here, is that if you treat UUCP as a full fledged domain then the temptation will be too great to dump everything destined for any UUCP site onto some unlucky overworked VAX somewhere and thus bottleneck the net to death while driving the admins to the poorhouse. Now unless I have this completely wrong, if I am sending to an address that lies in a domain of which I am also a member (say xyz.FOO.NET wants to send to bfmny0.UU.NET) then I am at liberty to try and parse the address (up to some limit, I scratch my head on this) within that known domain so as to optimize delivery. Whereas if it's a domain I'm not in, I should just send to the forwarder and let him deal with it. Now this is happening today by and large, for foreign nets am I right? If someone on MILNET wants to send me mail he has to go through the One True Gateway who does the forwarding... so gatewaying UUCP to Whatevernet is already relying on someone "stupid enough." However most UUCP mail incoming and outgoing (as of now) does have a UUCP site at the other end of the delivery path. So processing addresses *within* the domain could work as it does now with programs like smail. I'm sure I've overlooked something - my mailbox is open. :-) -- Tom Neff UUCP: ...!uunet!bfmny0!tneff "Truisms aren't everything." Internet: tneff@bfmny0.UU.NET
rsalz@bbn.com (Rich Salz) (06/05/89)
In article <56904@uunet.UU.NET> rick@uunet.UU.NET (Rick Adams) writes: >The trick is finding someone stupid enough to be willing to be the >uucp forwarder for 15,000 uucp hosts. In <14379@bfmny0.UUCP> tneff@bfmny0.UUCP (Tom Neff) writes: >Is there no way under the current/proposed scheme to share the >forwarding work load in an organized way? No. Under the domain system, you can basically say one of two things: 1. Give *.UUCP mail to me 2. Give a.UUCP, b.UUCP, c.UUCP, d.UUCP, etc., mail to me No one site is willing to forward for the world, so choice 1 is out. (Nor is it feasible for them to take in all the load then hand it off helper sites -- that's the same thing as routing to the final destination.) No five sites are going to maintain 3000 lines of data -- but it would be interesting to see if the current system could handle 15,000 flat-namespace records. :-) /rich $alz -- Please send comp.sources.unix-related mail to rsalz@uunet.uu.net. Use a domain-based address or give alternate paths, or you may lose out.
brad@looking.on.ca (Brad Templeton) (06/05/89)
Yes, there could easily be uucp.org with multiple forwarders, but that's not the point. There are many sites who simply stick ".uucp" on the end of their name. They would only be allowed to change to uucp.org once they had registered themselves with a forwarder. We would have the same problem. The old sites, with .uucp, that don't change, would still have to be routed by the old means. -- Brad Templeton, Looking Glass Software Ltd. -- Waterloo, Ontario 519/884-7473
rick@uunet.UU.NET (Rick Adams) (06/06/89)
> Is there no way under the current/proposed scheme to share the > forwarding work load in an organized way? You have just described the existing domain name system (e.g. foo.com etc) If you have to take the trouble to organize thigns, its trivial to fit into the existing domain heirarchy.
rick@uunet.UU.NET (Rick Adams) (06/06/89)
> No five sites are going to maintain 3000 lines of data -- but it would be > interesting to see if the current system could handle 15,000 > flat-namespace records. :-) It's called pathalias and it works fine now. UUNET (and lots of others) effectively do the hardpart already (e.g. the routing to foo.uucp to the final destination). The hard part is paying for the computer upgrades necessary to handle a huge increase in the mail volume once the entire 60,000 machine internet decides to route the mail through one site. --rick
zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff) (06/07/89)
>It's called pathalias and it works fine now. >The hard part is paying for the computer upgrades necessary to >handle a huge increase in the mail volume once the entire 60,000 machine >internet decides to route the mail through one site. > Let's say I create uucp.net and do name service for any *.uucp.net. I have some pathalias data that lists the route from various internet sites to individual uucp sites. I use this data for the MX request. Example: someone wants to send to foo.uucp.net. I find that rutgers talks to xxx which talks to foo. So rutgers gets the mail (it never arrives here). The net effect is that people can just use name.uucp.net if they are in the uucp maps without putting a huge load on one site. I'm not in favor of this, but it seems that it could be done. Something similar can be done now - try sending mail to user@any_uucp_name.uunet.net. It worked last time I tried it. -- Jon Zeeff zeeff@b-tech.ann-arbor.mi.us Ann Arbor, MI sharkey!b-tech!zeeff
rick@uunet.UU.NET (Rick Adams) (06/07/89)
> Let's say I create uucp.net and do name service for any *.uucp.net. I > have some pathalias data that lists the route from various internet > sites to individual uucp sites. I use this data for the MX request. But where do you get this data? Thats the hard part that everyone glosses over. > Something similar can be done now - try sending mail to > user@any_uucp_name.uunet.net. It worked last time I tried it. There are many things that are not currently blocked on uunet. As soon as the world starts dumping everything to uunet, I assure you that filtering will begin. Right now use of FOO.uu.net (not FOO.uunet.net) is strongly discouraged for non directly connected sites and if it get abused, it will be restricted. --rick
rynes@isis.CWRU.EDU (Edward M. Rynes Esq.) (06/08/89)
In article <57064@uunet.UU.NET> rick@uunet.UU.NET (Rick Adams) writes: >> Let's say I create uucp.net and do name service for any *.uucp.net. I >> have some pathalias data that lists the route from various internet >> sites to individual uucp sites. I use this data for the MX request. >But where do you get this data? Thats the hard part that everyone glosses >over. ********************************************************************** **** I sent this out once but it never made it off of our system **** ********************************************************************** In article <1780@fig.bbn.com> rsalz@bbn.com (Rich Salz) writes: >In article <56904@uunet.UU.NET> rick@uunet.UU.NET (Rick Adams) writes: >>The trick is finding someone stupid enough to be willing to be the >>uucp forwarder for 15,000 uucp hosts. >In <14379@bfmny0.UUCP> tneff@bfmny0.UUCP (Tom Neff) writes: >>Is there no way under the current/proposed scheme to share the >>forwarding work load in an organized way? >No. Under the domain system, you can basically say one of two things: > 1. Give *.UUCP mail to me > 2. Give a.UUCP, b.UUCP, c.UUCP, d.UUCP, etc., mail to me >No one site is willing to forward for the world, so choice 1 is out. There are a large number of UUCP/Internet gateways scattered around the world right now. The problem is that there is no way to get this information in a useful form. What I would like is for the UUCP Maps to start listing this information and for pathalias to start using it. This would help in two ways. First if we assume that all UUCP/Internet gateways can talk to each other directly (at very low cost via the Internet) then shorter, more direct paths can be generated by pathalias. For example, cwjcc and uunet are both Internet sites and can talk to each other directly. But pathalias doesn't know that so it generates the path: "mailrus!ames!mimsy!uunet!%s" And before I added mailrus to our maps it was much longer! Arpatext was supposed to solve this problem but it doesn't work. There are too many name collisions between Arpanet and UUCP hosts. This information can also be used to automatically generate a list of forwarders. The forwarder for any registered .uucp site could be the nearest registered UUCP/Internet gateway. If enough gateways are listed the load on each site would be small. Of course, if only a few gateways declare themselves then we are back to choice one above. :-( Of course this is all just mental gymnastics and not intended to reflect reality in any way. Use it as you see fit. :-) -- Edward M. Rynes <rynes@isis.CWRU.Edu> Jennings Computing Center "Hey!!! Somebody stole my quote!" Case Western Reserve University Cleveland, OH 44106 (216) 368-2982
erik@mpx2.mpx.com (Erik Murrey) (06/08/89)
In article <1780@fig.bbn.com> rsalz@bbn.com (Rich Salz) writes: >In article <56904@uunet.UU.NET> rick@uunet.UU.NET (Rick Adams) writes: >>The trick is finding someone stupid enough to be willing to be the >>uucp forwarder for 15,000 uucp hosts. > >In <14379@bfmny0.UUCP> tneff@bfmny0.UUCP (Tom Neff) writes: >>Is there no way under the current/proposed scheme to share the >>forwarding work load in an organized way? > >No. Under the domain system, you can basically say one of two things: > 1. Give *.UUCP mail to me > 2. Give a.UUCP, b.UUCP, c.UUCP, d.UUCP, etc., mail to me >No one site is willing to forward for the world, so choice 1 is out. >(Nor is it feasible for them to take in all the load then hand it off >helper sites -- that's the same thing as routing to the final destination.) This is why I think the .US domain is a *good thing*. If the coordinators can find enough internet forwarders, and divide the load appropriately, then .UUCP would(could) slowly go away. And since registration is free, I don't see why people wouldn't pass up the chance to have a proper domain name for their homes/school/businesses. If we had enough forwarders, in the key geographical areas, then your connection could be a local phone call away. As a small step, I will be willing to help out (time permitting) any internet site in setting up the proper sendmail/cf files to forward domains via UUCP. Please drop me a line via e-mail if you are interested. ... Erik -- Erik Murrey /| // /~~~~/ | / MPX Data Systems, Inc. / | / / /____/ |/ erik@mpx.com / / / / /| Data Systems, Inc. {vu-vlsi, bpa, cbmvax}!mpx1!erik / / / / |====================
david@dhw68k.cts.com (David H. Wolfskill) (06/08/89)
In article <56966@uunet.UU.NET> rick@uunet.UU.NET (Rick Adams) writes: >The hard part is paying for the computer upgrades necessary to >handle a huge increase in the mail volume once the entire 60,000 machine >internet decides to route the mail through one site. With all due respect (and appreciation!) for folks like Rick who run such a site (among a few other things...), as well as folks who design & build some amazingly reliable systems (both hardware and software) -- given the nature of what they're doing -- I would think that the notion of having the successful delivery of all that mail being dependent on one site might be a subject familiar to comp.risks readers.... I tend to be suspicious of a design that involves a single point of failure.... (This is not to say I refuse to use such a design; merely that I'm mildly paranoid about such things.... :-) Cheers, david -- David H. Wolfskill uucp: ...{spsd,zardoz,felix}!dhw68k!david InterNet: david@dhw68k.cts.com
zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff) (06/09/89)
In article <57064@uunet.UU.NET> rick@uunet.UU.NET (Rick Adams) writes: >> Let's say I create uucp.net and do name service for any *.uucp.net. I >> have some pathalias data that lists the route from various internet >> sites to individual uucp sites. I use this data for the MX request. > >But where do you get this data? Thats the hard part that everyone glosses >over. > From the uucp map entrys of the form "uucpname=name.domain". The hard part would be checking if these sites are willing and able to do the forwarding. -- Jon Zeeff zeeff@b-tech.ann-arbor.mi.us Ann Arbor, MI sharkey!b-tech!zeeff
brent@capmkt.COM (Brent Chapman) (06/10/89)
In article <9426@b-tech.ann-arbor.mi.us> zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff) writes: # In article <57064@uunet.UU.NET> rick@uunet.UU.NET (Rick Adams) writes: # >> Let's say I create uucp.net and do name service for any *.uucp.net. I # >> have some pathalias data that lists the route from various internet # >> sites to individual uucp sites. I use this data for the MX request. # > # >But where do you get this data? Thats the hard part that everyone glosses # >over. # > # # From the uucp map entrys of the form "uucpname=name.domain". The hard part # would be checking if these sites are willing and able to do the forwarding. When, oh, when will people learn that just because a site has a "domain style" name, that does _not_ mean it's on the InterNet. My site, for instance, is registered with the NIC as "capmkt.com", but we are most definitely _not_ an InterNet site; we're strictly UUCP-only, and thus totally unable to perform the type of gatewaying you propose. And even if we were able to do such forwarding, we probably wouldn't want to. This poor overloaded machine has more than enough to do without taking on the rest of the world's screwed up mail problems. -Brent -- Brent Chapman Capital Market Technology, Inc. Computer Operations Manager 1995 University Ave., Suite 390 brent@capmkt.com Berkeley, CA 94704 {apple,lll-tis,uunet}!capmkt!brent Phone: 415/540-6400
zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff) (06/12/89)
In article <300@capmkt.COM> brent@capmkt.UUCP (Brent Chapman) writes: >In article <9426@b-tech.ann-arbor.mi.us> zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff) writes: ># >> Let's say I create uucp.net and do name service for any *.uucp.net. I ># >> have some pathalias data that lists the route from various internet ># >> sites to individual uucp sites. I use this data for the MX request. ># > ># >But where do you get this data? Thats the hard part that everyone glosses ># >over. ># > ># ># From the uucp map entrys of the form "uucpname=name.domain". The hard part ># would be checking if these sites are willing and able to do the forwarding. > >When, oh, when will people learn that just because a site has a "domain >style" name, that does _not_ mean it's on the InterNet. My site, for As you should have been able to guess from "b-tech.ann-arbor.mi.us", I'm quite aware of this. You could either check each site to make sure it handles mail directly or simply allow the non internet site to do the forwarding. >definitely _not_ an InterNet site; we're strictly UUCP-only, and thus >totally unable to perform the type of gatewaying you propose. Why is a uucp site unable to do mail forwarding? Most do. >And even if we were able to do such forwarding, we probably wouldn't want >to. This poor overloaded machine has more than enough to do without taking >on the rest of the world's screwed up mail problems. Which is why I wrote: ># The hard part ># would be checking if these sites are willing and able to do the forwarding. A new entry in the uucp map data is probably the best way to handle this. -- Jon Zeeff zeeff@b-tech.ann-arbor.mi.us Ann Arbor, MI sharkey!b-tech!zeeff
allbery@ncoast.ORG (Brandon S. Allbery) (06/12/89)
As quoted from <9415@b-tech.ann-arbor.mi.us> by zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff): +--------------- | The net effect is that people can just use name.uucp.net if they are | in the uucp maps without putting a huge load on one site. I'm not | in favor of this, but it seems that it could be done. | | Something similar can be done now - try sending mail to | user@any_uucp_name.uunet.net. It worked last time I tried it. +--------------- Sorry, but *only* uunet subscribers are allowed to do that. ++Brandon -- Brandon S. Allbery, moderator of comp.sources.misc allbery@ncoast.org uunet!hal.cwru.edu!ncoast!allbery ncoast!allbery@hal.cwru.edu Send comp.sources.misc submissions to comp-sources-misc@<backbone> NCoast Public Access UN*X - (216) 781-6201, 300/1200/2400 baud, login: makeuser
mcb@ncis.tis.llnl.gov (Michael C. Berch) (06/12/89)
In <9436@b-tech.ann-arbor.mi.us> zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff)writes: > In article <300@capmkt.COM> brent@capmkt.UUCP (Brent Chapman) writes: > >When, oh, when will people learn that just because a site has a "domain > >style" name, that does _not_ mean it's on the InterNet. My site, for > > As you should have been able to guess from "b-tech.ann-arbor.mi.us", > I'm quite aware of this. You could either check each site to make > sure it handles mail directly or simply allow the non internet site to > do the forwarding. > > >definitely _not_ an InterNet site; we're strictly UUCP-only, and thus > >totally unable to perform the type of gatewaying you propose. > > Why is a uucp site unable to do mail forwarding? Most do. This totally misses the point of the previous discussion. "Forwarding", in this contest, MEANS "Internet forwarding". A UUCP-only site, by definition, is not on the Internet, and, ipso facto, is unable to do Internet forwarding. In order to do Internet forwarding you have to be ON the Internet, so that Internet hosts using the Domain Name System that receive MX records pointing at your host can contact you to send mail that is to be forwarded to the UUCP universe. It does not mean just "passing mail on". Mr. Zeeff, if you're having a problem getting through this, please drop me a note and I'll attempt to explain the best I can via e-mail. -- Michael C. Berch News/mail admin, tis.llnl.gov systems mcb@tis.llnl.gov / uunet!tis.llnl.gov!mcb
zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff) (06/12/89)
In article <233@ncis.tis.llnl.gov> mcb@ncis.tis.llnl.gov (Michael C. Berch) writes: >In <9436@b-tech.ann-arbor.mi.us> zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff)writes: >> In article <300@capmkt.COM> brent@capmkt.UUCP (Brent Chapman) writes: >> >When, oh, when will people learn that just because a site has a "domain >> >style" name, that does _not_ mean it's on the InterNet. My site, for >> >> As you should have been able to guess from "b-tech.ann-arbor.mi.us", >> I'm quite aware of this. You could either check each site to make >> sure it handles mail directly or simply allow the non internet site to >> do the forwarding. >> >> >definitely _not_ an InterNet site; we're strictly UUCP-only, and thus >> >totally unable to perform the type of gatewaying you propose. >> >> Why is a uucp site unable to do mail forwarding? Most do. > >This totally misses the point of the previous discussion. > "Forwarding", >in this contest, MEANS "Internet forwarding". A UUCP-only site, by definition, >is not on the Internet, and, ipso facto, is unable to do Internet forwarding. No, I meant forwarding as in any kind of forwarding that gets the mail there. Uucp only sites with domain names already have an internet forwarder that will get mail to them. These uucp sites can generally do additional forwarding, providing indirect "internet forwarding". I should have said "additional forwarding" instead of just "forwarding" to make this clear. >In order to do Internet forwarding you have to be ON the Internet, so >that Internet hosts using the Domain Name System that receive MX >records pointing at your host can contact you to send mail that is to >be forwarded to the UUCP universe. It does not mean just "passing mail on". See above. Assume uucp only site foo.com has an internet forwarder bar.edu and talks to another uucp site xxx. It may be perfectly reasonable for mail to travel bar.edu->foo.com->xxx. Unfortunately, the software most internet sites are using makes this impossible if xxx wants to use a domain name such as xxx.com and isn't a special case for bar.edu. xxx.uucp often works, perpetuating it's use. If you don't know someone at an internet site and can't afford a uunet connection, it is often difficult for a uucp site to use the DNS (which is why we may have .uucp forever). Solutions to this problem should be explored. Parts of a possible solution include 1) internet sites attempting to use the pathalias data when they find that mx records for some site point to themselves and 2) a way for such internet sites to register the fact that they are willing to provide uucp forwarding. I should probably be much verbose and fully explain the details to prevent these "you don't understand" postings. I'm off to usenix - send me email if you want to discuss this further. -- Jon Zeeff zeeff@b-tech.ann-arbor.mi.us Ann Arbor, MI sharkey!b-tech!zeeff
allbery@ncoast.ORG (Brandon S. Allbery) (06/16/89)
As quoted from <9436@b-tech.ann-arbor.mi.us> by zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff):
+---------------
| Why is a uucp site unable to do mail forwarding? Most do.
+---------------
In this context, "forwarding" means forwarding messages sent via SMTP over a
network from a site with an IP address to a site with an IP address; UUCP is
not and can not be integrated into such a network, it can only be layered on
top of such a network. That is, a random direct-connected Internet site can
not send mail to NCoast.ORG by having the nameserver return an [xx.yy.zz.ww]
IP address pointing to ncoast from an "A" record in the nameserver. We have
to have a forwarder; that is, we have an "MX" record in the nameserver which
gives the IP address (indirectly, via another lookup? I dunno, but it
doesn't really matter) of our forwarder; and our forwarder must be capable
of recognizing that even though the message was sent to them, it's really
for us.
++Brandon
--
Brandon S. Allbery, moderator of comp.sources.misc allbery@ncoast.org
uunet!hal.cwru.edu!ncoast!allbery ncoast!allbery@hal.cwru.edu
Send comp.sources.misc submissions to comp-sources-misc@<backbone>
NCoast Public Access UN*X - (216) 781-6201, 300/1200/2400 baud, login: makeuser
allbery@ncoast.ORG (Brandon S. Allbery) (06/16/89)
As quoted from <13756@ncoast.ORG> by allbery@ncoast.ORG (Brandon S. Allbery): +--------------- | In this context, "forwarding" means forwarding messages sent via SMTP over a | network from a site with an IP address to a site with an IP address; UUCP is +--------------- Uh, one other thing that may not be clear: the IP address is used by the network itself to indicate what site should grab the packet out of the network stream. TCP/IP is not store-and-forward, it transmits packets immediately, and the IP address states which system should pay attention to the packet. The packet recognition is done at too low a level (it might even be at the hardware level) to be "spoofed" after the fashion of MX records; the closest one can get to that is to tell the recognizer to accept anything with a subset of the IP fields having a specific value, and the subset must be consecutive items and start at the left -- which is how gateways to subnetworks are implemented. Thus, you can't just assign, say, NCoast.ORG an IP address to get around the restrictions on Internet mail. If you want to convince some direct Internet site to run TCP/SLIP on a modem for your benefit and install TCP/SLIP on your system, you can (in fact, must) get an IP address. But UUCP doesn't work like the IP protocol; it can't handle IP packets, and IP protocols can't handle UUCP packets, not just because the protocols have different packet configurations but because UUCP isn't designed to be running constantly and ignoring data not addressed to the local system. The two do not work in the same way at all. ++Brandon -- Brandon S. Allbery, moderator of comp.sources.misc allbery@ncoast.org uunet!hal.cwru.edu!ncoast!allbery ncoast!allbery@hal.cwru.edu Send comp.sources.misc submissions to comp-sources-misc@<backbone> NCoast Public Access UN*X - (216) 781-6201, 300/1200/2400 baud, login: makeuser
childers@avsd.UUCP (Richard Childers) (06/16/89)
mcb@ncis.tis.llnl.gov (Michael C. Berch) writes: >> Why is a uucp site unable to do mail forwarding? Most do. >This totally misses the point of the previous discussion. "Forwarding", >in this contest, MEANS "Internet forwarding". A UUCP-only site, by definition, >is not on the Internet, and, ipso facto, is unable to do Internet forwarding. >In order to do Internet forwarding you have to be ON the Internet, so >that Internet hosts using the Domain Name System that receive MX >records pointing at your host can contact you to send mail that is to >be forwarded to the UUCP universe. It does not mean just "passing mail on". What about playing with your sendmail.cf so that it recognizes Internet addresses and forwards them to an Internet site that _can_ forward them, in a classic Internet sense ? >Michael C. Berch >News/mail admin, tis.llnl.gov systems >mcb@tis.llnl.gov / uunet!tis.llnl.gov!mcb -- richard -- * "We must hang together, gentlemen ... else, we shall most assuredly * * hang separately." Benjamin Franklin, 1776 * * * * ..{amdahl|decwrl|octopus|pyramid|ucbvax}!avsd.UUCP!childers@tycho *
mcb@ncis.tis.llnl.gov (Michael C. Berch) (06/17/89)
In article <13757@ncoast.ORG> allbery@ncoast.ORG (Brandon S. Allbery) writes: > [...] Thus, you can't just assign, say, > NCoast.ORG an IP address to get around the restrictions on Internet mail. > > If you want to convince some direct Internet site to run TCP/SLIP on a > modem for your benefit and install TCP/SLIP on your system, you can (in > fact, must) get an IP address. But UUCP doesn't work like the IP protocol; > it can't handle IP packets, and IP protocols can't handle UUCP packets, not > just because the protocols have different packet configurations but because > UUCP isn't designed to be running constantly and ignoring data not addressed > to the local system. The two do not work in the same way at all. Uh, now *I'm* mixed up. I think Brandon is confusing a technical issue with a policy issue. There is no problem with using UUCP over a TCP/IP connection, whether it is SLIP or LAN or WAN or whatever. We do it all the time, using the TCP-UUCP code provided by Rick Adams (which I believe is distributed as the UUCP version in 4.xBSD). UUCP is just another TCP service, living on logical port 540. You make a connection and start up the protocol and there you are. The "restrictions on Internet mail", on the other hand, are policy based rather than technically based. There is absolutely no problem with getting a SLIP connection to someone on an Internet network, getting the NIC to assign you some IP namespace (i.e., a Class C network number), getting whomever runs the gateway for the site you connect to to advertise your net via EGP, and *poof*, you're on the Internet. You can do TCP-UUCP via that connection with anyone you please, but there's no reason to, since you can do SMTP with anyone you please, and that's much cleaner. The policy issue is permission to connect to the Internet. Mere possession of an official network number is not sufficient; the NIC will give those to anyone who asks. You also have to get permission from whomever your SLIP partner is connected to to put your packets on their net. In the old days, DOD had a monopoly on that, but this is no longer true. You will also have to negotiate about getting your connection known to the core gateways and may or may not need and Autonomous System Number, etc., etc. The technical issues are pretty trivial compared to the administrative/policy ones. -- Michael C. Berch mcb@tis.llnl.gov / uunet!tis.llnl.gov!mcb
bob@tinman.cis.ohio-state.edu (Bob Sutterfield) (06/17/89)
In article <13756@ncoast.ORG> allbery@ncoast.ORG (Brandon S. Allbery) writes:
...that is, we have an "MX" record in the nameserver which gives
the IP address (indirectly, via another lookup?...)
Nope, MX records can't point to other MXs. See RFC974.
sob@watson.tmc.edu (Stan Barber) (06/20/89)
In article <13757@ncoast.ORG> allbery@ncoast.ORG (Brandon S. Allbery) writes: >But UUCP doesn't work like the IP protocol; >it can't handle IP packets, and IP protocols can't handle UUCP packets, not >just because the protocols have different packet configurations but because >UUCP isn't designed to be running constantly and ignoring data not addressed >to the local system. The two do not work in the same way at all. > Not strictly true, Brandon. There are many sites that to UUCP over TCP/IP. For some sites, it's the only way they do TCP/IP. It's sorta like the experiments with RSCS that are being done with the BITNET II Project. The Internet provides what appears to be a leased-line for UUCP or RSCS to address the other site. It is correct that the design of UUCP is different that TCP/IP. STAN Stan internet: sob@bcm.tmc.edu Manager, Networking Olan uucp: {rutgers,mailrus}!bcm!sob Information Technology Barber Opinions expressed are only mine. Baylor College of Medicine
mcb@ncis.tis.llnl.gov (Michael C. Berch) (06/21/89)
In article <1320@avsd.UUCP> childers@avsd.UUCP (Richard Childers) writes: > mcb@ncis.tis.llnl.gov (Michael C. Berch) writes: > >This totally misses the point of the previous discussion. "Forwarding", > >in this context, MEANS "Internet forwarding". A UUCP-only site, by definition > >is not on the Internet, and, ipso facto, is unable to do Internet forwarding. > >In order to do Internet forwarding you have to be ON the Internet, so > >that Internet hosts using the Domain Name System that receive MX > >records pointing at your host can contact you to send mail that is to > >be forwarded to the UUCP universe. It does not mean just "passing mail on". > > What about playing with your sendmail.cf so that it recognizes Internet > addresses and forwards them to an Internet site that _can_ forward them, > in a classic Internet sense ? If you're talking about doing this on the UUCP side, that's presumably what UUCP-only sites are doing already in order to participate in the Domain Name System (specifically, the ability to mail to fully-qualified domain addresses that do not belong to one's UUCP neighbors.) This is an alternative to using the pathalias data from the UUCP map, but really should only be used to send to actual Internet sites, since otherwise the system may be used to go UUCP->Inet->UUCP, which violates the rules of *some* Internet component networks. The problem is that there's no way to tell, given a domain-type address, whether it is or is not on the Internet *without* having access to an Internet name server. I don't run a UUCP-only site, so don't have a sendmail.cf that does this. Someone around here undoubtedly does. All it really has to do is take any host address that has a dot in it where the domain isn't your own and isn't ".UUCP" and send it to a relay host which is on the Internet that can resolve arbitrary domain addresses. (Presumably this site is the one that you have arranged incoming Internet MX service with.) But this only solves the outgoing mail problem, not the incoming one. -- Michael C. Berch mcb@tis.llnl.gov / uunet!tis.llnl.gov!mcb