kagle@batcomputer.tn.cornell.edu (Jonathan C. Kagle) (04/16/87)
For the past month or so, I have been trying to get in touch with a friend at The University of St. Andrews, in Scotland. Unfortunately, the mail is swallowed somewhere along the way there. I can send mail to: username%uk.ac.st-and.sava@ac.uk with no positive (reply) or negative (error message) feedback from St. Andrews. I have tried to get through to the post- master/mistress, but I have recieved no reply. I have already sent mail to bogus addresses, with varying success. I sent a letter to postmaster%uk.ac.st-and.xxx@ac.uk, and other permutations, and I get a "No Such Address"-type error message from the BITNET-JANET gateway. However if I send a message to xxx%uk.ac.st-and.sava@ac.uk, I get no response at all (BTW, I did try several replacements for the xxx's, so there could be no chance that I reached real usernames at St. Andrews.). Could someone help me get through? Have I found a mailer bug? Should I use another gateway? HELP! -Jonathan C. Kagle
bct@its63b.ed.ac.uk (B Tompsett) (04/17/87)
In article <701@batcomputer.tn.cornell.edu> kagle@tcgould.tn.cornell.edu.UUCP (Jonathan C. Kagle) writes: > > For the past month or so, I have been trying to get in touch with a >friend at The University of St. Andrews, in Scotland. Unfortunately, the >mail is swallowed somewhere along the way there. [...] > [...] I can send mail to: >username%uk.ac.st-and.sava@ac.uk with no positive (reply) or negative (error >message) feedback from St. Andrews. I have tried to get through to the post- >master/mistress, but I have recieved no reply. > [...] > Could someone help me get through? Have I found a mailer bug? Should >I use another gateway? HELP! You are asking questions in a sensitive area. There has been some heated exchanges on this subject at this end; This reply is therefore worded somewhat carefully in the light of recent events. I have recently returned to the UK from the States and therefore have experienced the problems of the kind you describe from both ends. There are three main gateways into the UK, (and they are all described in the CACM article on Notable Networks), and as far as I know, each of them has to pay bi-directionally on transatlantic mail. Therefore each of your attempts to mail to St. Andrews was billed to the gateway, or St. Andrews. When the network in the UK was first set up funding was available and also the common carrier made its facilities available free/cheap. Now this has changed, and real money is paid to send and recieve mail/news and at the same time the Universities and other institutions are feeling the pinch of austerity measures caused by the paymasters wishing to spend their money elsewhere. As a result some sites decided not to subscribe to all the gateways. As the gateways themselves also had the same funding problems they had to make some policy decisions on what to do with mail arriving for a site that would no longer pay to receive its mail. To return the message would mean that the gateway would have to pay to handle the mail twice and recieve no re-imbursement; they therefore decided to just forward it to /dev/null. I am not claiming that this is what has happened with your mail, that depends on the current relationship between your recipient, his host site, and the gateway; I have no specific information in your case. I would like to offer advice to you on the best routing that would enable your message to get through, but current circumstances make that impossible in a public forum. It would be regarded as an attempt to yell "fire" in a crowded theatre. Particularly when in this case there are not many exits from the theatre. I will however endeavor to mail you directly with some helpful advice, unless I am strongly advised not to do so by local gateway authorities. It should therefore be concluded that there is no reliable routing path from the US to the UK for unsolicited mail. If your mail is of particular importance your site should make arrangements to connect directly with any destination site in the UK using X.25 over IPSS. All that remains to be done is to agree on the protocol to be used. (In other words, should your site send packets in UK format, or should your recipient understand your sites protocol?) With regard to trying to contact the Postmaster, note that a recent survey of JANET sites showed that only 51% (330 of 642) of Postmasters at registered sites would respond to requests. Some sites do not even implement Postmaster despite RFC822 and a similar UK standard. I know for a fact that many of the systems at my site do not have an active Postmaster, and mail just builds up for Postmaster, root and the like. The administrators claim they have better things to do than handle silly mail problems (Their words). I hope this description of the current state of affairs helps put you in the picture, and also educates a wider audience at the same time. Disclaimers: This posting represents the views of the writer and do not form an official opinion of his employers, or the network administrators at this site or any other UK site or UK gateway. No critisism of the efforts of the network staff or administrators at any site or gateway is intended. They are doing a fine job in difficult circumstances. -- -- > Brian Tompsett. Department of Computer Science, University of Edinburgh, > JCMB, The King's Buildings, Mayfield Road, EDINBURGH, EH9 3JZ, Scotland, U.K. > Telephone: +44 31 667 1081 x3332. > JANET: bct@uk.ac.ed.ecsvax ARPA: bct%ecsvax.ed.ac.uk@cs.ucl.ac.uk > USENET: bct@ecsvax.ed.ac.uk UUCP: ...!seismo!mcvax!ukc!ecsvax.ed.ac.uk!bct > BITNET: psuvax1!ecsvax.ed.ac.uk!bct or bct%ecsvax.ed.ac.uk@earn.rl.ac.uk
jim@cs.strath.ac.uk (Jim Reid) (04/24/87)
In article <701@batcomputer.tn.cornell.edu> kagle@tcgould.tn.cornell.edu.UUCP (Jonathan C. Kagle) writes: > > For the past month or so, I have been trying to get in touch with a >friend at The University of St. Andrews, in Scotland. Unfortunately, the >mail is swallowed somewhere along the way there. I can send mail to: >username%uk.ac.st-and.sava@ac.uk with no positive (reply) or negative (error >message) feedback from St. Andrews. I have tried to get through to the post- >master/mistress, but I have recieved no reply. > I have already sent mail to bogus addresses, with varying success. I >sent a letter to postmaster%uk.ac.st-and.xxx@ac.uk, and other permutations, and >I get a "No Such Address"-type error message from the BITNET-JANET gateway. >However if I send a message to xxx%uk.ac.st-and.sava@ac.uk, I get no response >at all (BTW, I did try several replacements for the xxx's, so there could be >no chance that I reached real usernames at St. Andrews.). There are three main gateways into the UK. UKC have the uucp gateway, UCL the arpa/csnet/x400 gateways and RAL have a gateway into BITNET/EARN. For reasons of funding, neither UKC or UCL will forward mail for unauthorised users. In the UK we have to pay the network bills for transatlantic traffic, so if someone can't or won't pay their share of these costs, UCL and UKC have not much option but to silently junk the mail. [This is not as bad as it seems; all the UK UUCP sites are registered with UKC and pay up. There would be no problem there, except the site you want to get to is not a UUCP site - it's a VMS VAX. It's likely that if your mail ends up at UCL or UKC it goes to /dev/null and you'll hear nothing about that fact.] Now the BITNET/EARN gateway is currently free - I think IBM have sponsored a line till 1988. There shouldn't be any problem of authorisation there, at least for now. The address you give is a little curious. You are asking for the mail to be sent to "ac.uk" from your local machine. Where will your mail system send this (if anywhere)? If it goes to the RAL gateway - called UKACRL on BITNET/EARN - it should get through OK. The UK address you give is more or less correct, but could throw the gateway. Firstly the address is in UK domain order. The RAL gateway (or some mailer inbetween you and it) could swap the order round to RFC822 form. This could upset the gateway if it expects addresses to be in one of either RFC822 or UK order and gets presented with the other order. Secondly, the UK academic network has two names for each site (and you thought swapping the domain ordering was bad...)! These names are known as long and short form names, the long form names being the standard "approved" ones. The gateway at RAL could be fussy enough to only accept the standard form name. The address you gave - uk.ac.st-and.sava - is the short form name. The standard form name is uk.ac.st-andrews.sava. The gateways at UKC and UCL are clever enough to swap domain ordering around and map between long and short names. In short, they hide the lunacy of the UK naming scheme from the rest of the world. I don't know how well the RAL gateway copes with this, if at all. Ask the postmaster there. My advice to you is first determine which gateway you used (or want to use) and ask the postmaster there how mail addresses for the UK can be presented to his/her mail system. From what you said, it sounds like your mail didn't always make it to a gateway, so I'd suggest you trace what happens on the path to the gateway and back to you. I don't think a UK gateway is at fault - they are quite good at returning error/status messages, so I would suspect your path to the gateway is not reliable. If the gateway is forwarding the mail OK, the problem must then be at St. Andrews. Perhaps your friend there isn't able to send mail back to you. It is also possible that there is no postmaster there - about 20-30% of sites in JANET don't even have a postmaster! (JANET is the UK Academic network.) > Could someone help me get through? Have I found a mailer bug? Should >I use another gateway? HELP! Well, I think I've given enough explanation above. It's not a mailer bug, though someone's mailer could well be misconfigured. Please mail me (if you can) for more information. I'd be glad to help. I felt that a net-wide posting could be of general benefit to others in the same position as you. Jim PS: Can you imagine the fun we had when someone in the UK wanted to register uk.ac.xxxx.com? PPS: 256 -rw------- 1 root 256112 Jan 26 16:10 /usr/lib/sendmail.fc Is this a record? ARPA: jim%cs.strath.ac.uk@ucl-cs.arpa, jim@cs.strath.ac.uk UUCP: jim@strath-cs.uucp, ...!seismo!mcvax!ukc!strath-cs!jim JANET: jim@uk.ac.strath.cs "JANET domain ordering is swapped around so's there'd be some use for rev(1)!"
dik@mcvax.cwi.nl (Dik T. Winter) (04/25/87)
In article <572@stracs.cs.strath.ac.uk> jim@cs.strath.ac.uk writes: > In article <701@batcomputer.tn.cornell.edu> kagle@tcgould.tn.cornell.edu.UUCP (Jonathan C. Kagle) writes: > > > > For the past month or so, I have been trying to get in touch with a > >friend at The University of St. Andrews, in Scotland. Unfortunately, the > >mail is swallowed somewhere along the way there. I can send mail to: > >username%uk.ac.st-and.sava@ac.uk with no positive (reply) or negative (error > >message) feedback from St. Andrews. I have tried to get through to the post- > >master/mistress, but I have recieved no reply. > > There are three main gateways into the UK. UKC have the uucp gateway, UCL > the arpa/csnet/x400 gateways and RAL have a gateway into BITNET/EARN. For > reasons of funding, neither UKC or UCL will forward mail for unauthorised > users. In the UK we have to pay the network bills for transatlantic traffic, (Same holds here, so we pay for reading the original message as for for sending this reply, but we do not complain.) > so if someone can't or won't pay their share of these costs, UCL and UKC have > not much option but to silently junk the mail. [This is not as bad as it seems; > all the UK UUCP sites are registered with UKC and pay up. There would be no > problem there, except the site you want to get to is not a UUCP site - it's > a VMS VAX. It's likely that if your mail ends up at UCL or UKC it goes to > /dev/null and you'll hear nothing about that fact.] > > Now the BITNET/EARN gateway is currently free - I think IBM have sponsored a > line till 1988. There shouldn't be any problem of authorisation there, at > least for now. > > The address you give is a little curious. You are asking for the mail to > be sent to "ac.uk" from your local machine. Where will your mail system > send this (if anywhere)? If it goes to the RAL gateway - called UKACRL > on BITNET/EARN - it should get through OK. > > The UK address you give is more or less correct, but could throw the gateway. > Firstly the address is in UK domain order. The RAL gateway (or some mailer > inbetween you and it) could swap the order round to RFC822 form. This could > upset the gateway if it expects addresses to be in one of either RFC822 or > UK order and gets presented with the other order. Secondly, the UK academic > network has two names for each site (and you thought swapping the domain > ordering was bad...)! These names are known as long and short form names, the > long form names being the standard "approved" ones. The gateway at RAL could > be fussy enough to only accept the standard form name. The address you gave > - uk.ac.st-and.sava - is the short form name. The standard form name is > uk.ac.st-andrews.sava. > > The gateways at UKC and UCL are clever enough to swap domain ordering around > and map between long and short names. In short, they hide the lunacy of the > UK naming scheme from the rest of the world. I don't know how well the RAL > gateway copes with this, if at all. Ask the postmaster there. As far as I know, the RAL gateway does also deal with these problems, although I think the preference is in the non-uk order (i.e. terminating with .ac.uk). > > My advice to you is first determine which gateway you used (or want to use) > and ask the postmaster there how mail addresses for the UK can be presented > to his/her mail system. From what you said, it sounds like your mail didn't > always make it to a gateway, so I'd suggest you trace what happens on the > path to the gateway and back to you. I think the mail may have gone to a gateway, but the wrong one. The EARN site (European part of bitnet) I just consulted tells me that mail to Janet is still gatewayed through UCL, although this is false (it is gatewayed through RAL). Now it may be that one of the intermediate sites still gateways it through UCL. You may have to send it to RAL by hand with a properly formatted mail message, although I do not know right know what that format is. I have ways to look it up though, but it may take some time. You might also ask the postmaster (that should be postmaster@ukacrl.bitnet). > > I don't think a UK gateway is at fault - they are quite good at returning > error/status messages, so I would suspect your path to the gateway is not > reliable. No, at least UKC will not return any error/status message, and I think UCL won't either. -- dik t. winter, cwi, amsterdam, nederland INTERNET : dik@cwi.nl BITNET/EARN: dik@mcvax
zben@umd5.umd.edu (Ben Cranston) (04/25/87)
In article <572@stracs.cs.strath.ac.uk> jim@cs.strath.ac.uk writes: > The address you give is a little curious. You are asking for the mail to > be sent to "ac.uk" from your local machine. Where will your mail system > send this (if anywhere)? If it goes to the RAL gateway - called UKACRL > on BITNET/EARN - it should get through OK. The semistandard BitNet domain route table has "ac.uk" pointed to ukacrl. This makes life somewhat difficult for those of us who try to provide both the Internet and BitNet addressing domains... > Firstly the address is in UK domain order... The UKACRL router does not seem to reverse the domain order, unlike the other two gateways, thus presentation in UK order seems correct here... -- umd5.UUCP <= {seismo!mimsy,ihnp4!rlgvax}!cvl!umd5!zben Ben Cranston zben @ umd2.UMD.EDU Kingdom of Merryland UniSys 1100/92 umd2.BITNET "via HASP with RSCS"
henry@garp.mit.edu (Henry Mensch) (04/26/87)
In article <572@stracs.cs.strath.ac.uk> jim@cs.strath.ac.uk writes:
For reasons of funding, neither UKC or UCL will forward mail for unauthorised
users. In the UK we have to pay the network bills for transatlantic traffic,
so if someone can't or won't pay their share of these costs, UCL and UKC have
not much option but to silently junk the mail.
Well, ucl *doesn't* silently junk mail addressed to unauthorised users;
my experience shows that cs.ucl.ac.uk returns the mail to the sender
with a message stating that it doesn't gratuitously forward mail.
#
# Henry Mensch / <henry@garp.mit.edu> / E40-358C MIT, Cambridge, MA
# {ames,cca,rochester,mit-eddie}!garp!henrykre@munnari.oz (Robert Elz) (04/26/87)
In article <1571@umd5.umd.edu>, zben@umd5.umd.edu (Ben Cranston) writes: > The semistandard BitNet domain route table has "ac.uk" pointed to ukacrl. > This makes life somewhat difficult for those of us who try to provide > both the Internet and BitNet addressing domains... Yes, this is a disaster, though by assuming that anything from a host masquerading as "ac.uk" (full domain name) is really from ukacrl.bitnet its possible to restore some sense to the universe. What's worse is that they set the return-path on mail they gateway from JANET to the rest of the world to "EMAILDEV@UKACRL.BITNET" (I guess the ".BITNET" isn't there while still in bitnet itself). That means that anyone in the UK who sends mail out through that gateway, and sends it to an bad address (for whatever reason) is never going to get their message back. It used to be true that there wasn't even a mailbox at that address, so that any mail bounced back to it, was just returned to the site that originally bounced it. I'd hate to count how much mail I received because of this (ie: postmaster@munnari.oz.au bounces mail because of a bad Australian address, then Rutherford bounce it back to postmaster@munnari.oz.au (me) because EMAILDEV doesn't exist). I haven't seen any of this kind of double bounced mail for a while now, so its possible this is finally fixed. Someone might even forward on the bouncing mail to the correct recipient. But in general, I would advise everyone not on bitnet to avoid that gateway, its clearly inferior to both UCL and UKC by several orders of magnitude. What's more it means that your mail goes through the tortuous transformations needed to get on and off bitnet, for no particularly good reason. Robert Elz kre@munnari.oz.au
zben@umd5.umd.edu (Ben Cranston) (04/27/87)
In article <1582@munnari.oz> kre@munnari.oz (Robert Elz) writes: > In article <1571@umd5.umd.edu>, zben@umd5.umd.edu (Ben Cranston) writes: >> The semistandard BitNet domain route table has "ac.uk" pointed to ukacrl. >> This makes life somewhat difficult for those of us who try to provide >> both the Internet and BitNet addressing domains... > Yes, this is a disaster, though by assuming that anything from a host > masquerading as "ac.uk" (full domain name) is really from ukacrl.bitnet > its possible to restore some sense to the universe. I assume the "full domain name" means to transform "ac.uk" to "ukacrl" but leave "<anything>.ac.uk" alone. This would probably work. My local problem is use of somebody else's code and database for site lookup, and the said somebody else not wanting to include special cases like this. On balance, I'm not sure I disagree with him... > What's worse is that they set the return-path on mail they gateway > from JANET to the rest of the world to "EMAILDEV@UKACRL.BITNET" (I > guess the ".BITNET" isn't there while still in bitnet itself). > That means that anyone in the UK who sends mail out through that > gateway, and sends it to an bad address (for whatever reason) is > never going to get their message back. I guess you mean the RSCS out-of-band information is EMAILDEV and UKACRL. I wasn't aware that one could arbitrarily set these fields, due to system security considerations. That is, I had always assumed there was a virtual machine called EMAILDEV that routed the mail. In any case, if a mail relay relied upon the RSCS out-of-band information to create the SMTP "MAIL FROM:" data, the situation you describe would occur. After some thought I coded my relay to attempt to extract the information from the "From:" (or "Sender:") information in the header, and to ignore the RSCS information (which is also something like SMTPUSER WISCVM for mail that has come through the WISCVM gateway). So now when I process mail from the UKACRL gateway it is erroniously marked as "MAIL FROM:<luser%xx.yy.zz@AC.UK>" instead of being erroniously marked as "MAIL FROM:<EMAILDEV@UKACRL.BITNET>" ... fooey. Some days it doesn't even pay to get out of bed. > But in general, I would advise everyone not on bitnet to avoid that gateway, > its clearly inferior to both UCL and UKC by several orders of magnitude. > What's more it means that your mail goes through the tortuous transformations > needed to get on and off bitnet, for no particularly good reason. If people won't play by the rules, there's not much a poor postmaster can do except to forewarn the innocent. -- umd5.UUCP <= {seismo!mimsy,ihnp4!rlgvax}!cvl!umd5!zben Ben Cranston zben @ umd2.UMD.EDU Kingdom of Merryland UniSys 1100/92 umd2.BITNET "via HASP with RSCS"
kre@munnari.oz (Robert Elz) (04/27/87)
In article <1574@umd5.umd.edu>, zben@umd5.umd.edu (Ben Cranston) writes: > I assume the "full domain name" means to transform "ac.uk" to "ukacrl" but > leave "<anything>.ac.uk" alone. This would probably work. It seems to. > My local problem is use of somebody else's code and database for site > lookup, and the said somebody else not wanting to include special cases > like this. On balance, I'm not sure I disagree with him... I don't really either, but its a case of making mail work. The right people to fix this are Rutherford Labs, but they don't seem to be in any hurry to do anything about it. > I guess you mean the RSCS out-of-band information is EMAILDEV and UKACRL. No, not really. I know nothing about RSCS or BITNET really, but I thought that generally mail was transferred using BSMTP. I had assumed that BSMTP would have a "Mail From:<xxx>" in its envelope. That was the one I was suggesting that Rutherford should set. If BSMTP doesn't have this header, perhaps it should? If Rutherford don't use BSMTP then perhaps they should? Robert Elz kre@munnari.oz.au
pc@eagle.ukc.ac.uk (R.P.A.Collinson) (04/28/87)
Some folk-lore instantly springs into existence at every one of the mailings under this heading. Can I dispel some please. a) We run exactly the same mail system as UCL. If we receive a message which is badly addressed or fails for any other reason in the mailer, we attempt to return it to the sender. To save cost, we now truncate the message. I expect that a number of these returned messages are lost - but then this will always be the case when you consider the general difficulty of asking your local mail program to `reply' to mail. Please mail to uknet@ukc if you are experiencing any problem. b) We currently attempt to deliver all correctly addressed mail inbound to sites in the UK, we don't like to do this to sites which do not appear in the UUCP maps for the UK because it incurs expenditure which cannot be passed onto the end user's site. Outbound mail is constrained by an authorisation system using the name of the UK site. We will only forward mail FROM the sites in the UUCP maps for the UK, this CAN mean the loss of an error response emitted by an unregistered site to you. c) Assuming that the mail was coming into the UK via ukc, then the problem with the St. Andrews VAX was, I think due to the FTP mechanism (not the US FTP, I hasten to add - but the UK `Blue Book FTP' system). FTP is supplied with addresses from a central register and it seemed reasonable that if a name appeared there saying that the machine would/could get mail, then FTP would keep trying in the fervent belief that someone someday would wake up to the fact that there was nothing coming in. We have altered this policy and now operate a 3 week bounce time on mail (this is perhaps a too long, but it can be tuned and that is what we may do). (I think that the case in point MIGHT have been due to a mismatch between our FTP system and that running on VMS machines - ie a `bug'). d) We accept addresses in any domain order and swap them to the `correct' order for the appropriate network. I believe that the UK Bitnet gateway is planning to do this in June or perhaps July. e) I should stop, the yeast for my bread is erupting out of the cup.