MINUTES Nordunet Mail Inspector session Time: Wednesday October 2nd, 0930 - 1145. Place: UNI-C, Vermundsgt. 5, Copenhagen, Denmark. Participants: Alf Hansen Alf.Hansen@delab.sintef.no Jan P. S|rensen Jan.P.Sorensen@uni-c.dk Odd Asbj|rn Halseth Odd.A.Halseth@delab.sintef.no Per Andersson pa@cdg.chalmers.se Harri Salminen hks@funet.fi Svante Eriksson, Swedish Telecom Intenational Marko Kaittola Marko.Kaittola@funet.fi Manu Mahonen Manu.Mahonen@funet.fi Petri Ojala ojala@funet.fi Jyrki Soini Jyrki.Soini@funet.fi Erik Lawaetz Erik.Lawaetz@uni-c.dk Roland Hedberg Roland.Hedberg@umdac.umu.se Klaus Hansen Khan@diku.dk Harald Tveit Alvestrand Harald.Alvestrand@delab.sintef.no 0) Introduction by Harald Tveit Alvestrand Harald introduced the NORDUnet Mail Inspector (NMI), Alf Hansen, who chaired this session, which was the first meeting between the NMI and the national E-mail managers in NORDUnet. 1) Introduction and round table presentation The distributed agenda was accepted with no comments. 2) Definition of the NORDUnet MHS Service There was a discussion of this point, which seemed to come down pretty quickly to the conclusion below. Things mentioned during the discussion: Marko Kaittola: We do not need to worry about toplevel domain .uucp. EUnet will drop toplevel domain .uucp and support for bang addresses at end of 91. Jens P: We would have to register mappings for all the users that we want to communicate with. It was pointed out that discussions on mapping strategies between X.400 and RFC-822 are going on on the ietf-osi-x400ops@cs.wisc.edu list, and that we could not solve all these problems in this meeting. Is it a goal to harmonize address space with the non-R&D community? Yes, probably. At least we in NORDUnet should have this as a goal and try to reach it. The non-academic X.400 services are outside our control, but we want to have good relations with them, allowing address harmonization here also. Manu: Problems with integrating with the HEPNET/SPAN community, which uses DECNET mail. CONCLUSIONS: - Limit the NORDUnet MHS service to those users who have RFC-822 addresses with legally registered toplevel domains and X.400 Standard Attribute addresses within the Nordic countries - Have as a goal to coordinate address spaces with the non-R&D communities in the Nordic countries - One goal is to come up with a mapping scheme usable in all Nordic countries for mapping non-academic X.400 domains into the RFC-822 world. 3) Identification of national contacts Discussion about who should be national contacts - nations, networks or services? It was decided to try to list contacts with *all* the groups that offer E-mail services, and not have anyone representing groups unless they had some kind of formal connection with them. EXAMPLE LIST (to be worked on): Dennmark (DK): - Domain registered by DKUUG DKnet (the DKUUG network) Kim Storm R&D MHS Klaus Hansen EARN Jan P. S|rensen,UNI-C Danish DECnet ditto Public service providers - 3 ADMDs. ** ACTION DENNMARK: Sort out who the contact persons and their E-mail addresses are, before end of next week. Sweden (SE): - Bj|rn Eriksen is registering; unclear wherther SUNET or EurOpen-S actually "own" the domain; he may represent both. SUNET Anders Gillen or Bj|rn Eriksen EurOpen-S (Swedish UUCP) Bj|rn Eriksen SWIPnet (commercial TCP/IP) ? TIPnet (Televerket commercial IP) (small yet) SUNET X.400 Per Andersson TeleDelta (Commercial X.400 ADMD) Svante Eriksson? ** ACTION SWEDEN: Sort out names and their E-mail addresses before end of next week. Finland (FI) R&D MHS Marko EUNET Petri DECNET/BITNET Mano Commercial ADMDs Go via Marko Memonet too early to say ** ACTION FINLAND: Sort out missing names and their E-mail addresses before the end of next week. Norway (NO) R&D MHS Harald Tveit Alvestrand NUUG Hans Petter Holen TelemaX.400 will find a name ** ACTION NORWAY: Sort out names and their E-mail addresses before the end of next week. Iceland (IS) Nobody from Iceland was present. The NMI will set up a list containing *only* the above mail managers, in order to have a target for more "high priority" messages, like "take actions". The name of the list shoud be mhs-managers@nic.nordu.net 4) Definition of service goals for NORDUnet MHS Examples of goals: - postmaster for all organizations - request address for all mailing lists - coordinated use of From:, Sender: on all distributionlist (quality of DL service) - consistent handling of error messages (who do we send to) - content of error messages - how do we understand them? - input of addresses in all mail systems as written on a card in standard format - too ambitious? - all messages should be repliable - backup MTAs for machines that are often down - do not use hostname addresses to machines that are turned off when people go home - people should have stable, non-machine addresses - use . and not _ or / as separator between Givenname and Surname in the RFC-822 representation of the address - maximum delay enduser to enduser (f.ex 90% within 5 min) - cooperation with other countries' services - Meet or beat COSINE service requirements? Note that things like average processing time through gateways came a long way down the list; it is not this that is the problem! The discussion will continue on the list, and the NMI will make a suggestion for a set of QOS goals. ** ACTION NMI: To propose a list of service goal parameters with values to be accepted by the MHS-Managers at the next meeting. 5) Measurements Very hard to measure mail that goes directly from originator's workstation to recipient's workstation. Only possibility is eavesdropping devices on the backbone networks. To us, the service managers, the gateways are important, and can be checked by using "echo" scripts or similar things on a regular basis (Harri). We should distinguish between error reporting (what happens when the user does not get his mail through) and samping (what happens when we try the route using some tool) (Klaus Hansen). We should collect and publish a list of echo addresses, for use in connectivity testing. Traffic statistics should perhaps not be done always, but sometimes, to identify the "busy links", in order to plan the monitoring. (example given from the Copenhagen bus service's scheme) It would be nice if the NMI could collect programs that are used to collect the statistics. Common-log-format formatter? ** ACTION ALL GATEWAY MANAGERS: Send statistics (whatever is available) to the Mail Inspector. Address: NMI@nic.nordu.net 7) Next Meeting - Meetings should only be when we have CONCRETE things to discuss (Klaus) - A meeting to evaluate a result pushes us to do something (Marko) - Not an advantage together with NETF - too many other things (Marko) - Problem with adding YET another day of travel, as is the case when it is not at the NETF SUGGESTED NEXT MEETING: At the NETF, Wednesday, March 4, 1992. 6) Brainstorming on Nordunet Mail Harmonization Workplan This workplan is a *result* from the 1991/92 year of NMI, and gives the work to be done in the *next* period. Will be taken up again in March; too little basis here. 8) Any Other Business. No other business. MINUTES NETF-MHS group, meeting about Internet Mail Extensions Time: Wednesday October 2nd, 1500-1730 Place: UNI-C, Vermundsgt. 5, Copenhagen, Denmark. Minutes taken by Peter Hausken Present: Patrik Faltstrom pat@nada.kth.se Per Andersson pa@cdg.chalmers.se Johnny Eriksson bygg@sunet.se Erik Lawaetz Erik.Lawaetz@uni-c.dk Marko Kaittola Marko.Kaittola@funet.fi Matti Aarnio Matti.Aarnio@funet.fi Manu Mahonen Manu.Mahonen@funet.fi Petri Ojala ojala@funet.fi Jyrki Soini Jyrki.Soini@funet.fi Pekka Kystolaakso Pekka.Kystolaakso@csc.fi B Svante Erikkson, ST International Bengt Ackzell G=bengt;S=Ackzell;O=televerket; ADMD=tele;C=se Stig Ouvrier Stig.Ouvrier@televerket.data.se (G=Stig;S=Ouvrier;O=Televerket; ADMD=tede;C=se) Jan P. Sorensen Jan.P.Sorensen@uni-c.dk Keld Simonsen Keld.Simonsen@dkuug.dk Peter Hausken Peter.Hausken@usit.uio.no Alf Hansen Alf.Hansen@delab.sintef.no Harald gave a summary of current status of RFC-XXXX (X4) Keld Simonsen gave a summary of the status of his RFCs on charactersets The main question: Is this usable for NORDUnet? Notes from the discussion: - Get it into the RFC where EDI should be placed. Exaples of how EDI could (should) be encoded. - The RFC-XXXX only covers 7-bit transmission of SMTP. 8-bit is not covered. - Why this extension? X.400 88 got the same. Reply: SMTP is an existing infrastructure. X.400 88 is not out in products. - How is binary data treated acording to the 1000 character line limit? Reply: Binary should always be encoded BASE64 or Quoted-Printable. - Keld Simonsen: It`s good for the body part. We might have to do something with the header. - It might be pushed upon us anyway from the vendors if the RFC gets aceptance in the US. - What should be used for encoding 8-bit characters? Quoted-Printable or Mnemonic? Mnemonic is more readable to the user. Quoted-Printable are easier to implement. The Mnemonic code is in the public domain available at: dkuug.dk:/pub/ch.shar01 og ch.shar02 Bj|rn Aczell of the Swedish PTT thought that it might be difficult to push vendors into accepting "yet another nonstandard coding scheme"; other participants wondered about the value of the mail products delivered by the vendors with their systems. A brief debate on the use of the Nordic 7-bit variants ({|} used for national characters) indicated the presence of people who were completely for it and people who were altogether against it. No consensus there. (it is possible to indicate the use of such a charset in the RFC-XXXX headers) Conclusions: - We want to see RFC-XXXX used wihin NORDUnet - We see need of further work on encoding headers - Intermidiate MTAs should in general not change the message unless necesary (through gateways oe.) - When encoding characters, the prefered scheme is Mnemonic, but others should be handled. - We do not regard 8-bit transport as essiensial, but it might be useful if it's well-defined. - Mail Useragents should at least represent the ISO-8859-1 characterset reporoire correctly as far as the hardware allows. Two pieces that we felt necessary which are missing now, are: - Use of national characters in the header fields - Gatewaying of RFC-XXXX messages to X.400-based systems (coding it as IA5text is QUITE user-unfriendly) Patrik F{ltstr|m gave a presentation on his suggestion for a possible extension of headers to allow national characters in the header. (I have left this with the octal codes....) To: paf@nada.kth.se (P. F\204ltstr\224m) Subject: From: Patrik F\204ltstr\224m Header-CharacterSet: ISO-8859-1 | US-ASCII Header-transport-encoding: Quoted-Printable | Mnemonic This is surprisingly similar to the result of the discussion we had at NETF in Stockholm. The idea was killed off then on the IETF mailing list, but nobody remembers exactly why. Objections against parsing the headers two times; there seems no way to avoid this as long as MTAs can scramble the headers any way they like, so that we cannot guarantee that the Header- fields will be first. (Someone claimed that Sendmail actually scrambles headers) Conclusion: - If we can get national characters we would like it, but if it stops RFC-XXXX we wait. The national characters in the body is more important than in the header! Some people wondered whether a "tack-on" RFC describing headers would be implemented; we will talk more about tactics on this later. How do we we get there? We need MTAs and User Agents + gateways supporting RFC-XXXX It`s not only the coding needed, but also support and bug fixes. Funds may possibly be available through NORDUNET or other mechanisms. Names that were brought up: Jan Sell, Sverige (ELM) Jan Engvall, Lund (implemented MacPost) Some people in Computer Science, ]rhus (known to Keld Simonsen) Matti Aarnio Steve Kille (probably working on gatewaying it) As well we need the same for NetNews readers. Harri Salminen mentioned that the IETF group working on News extensions wanted to simply copy the work of RFC-XXXX rather than try to invent something different. Good! MINUTES NETF MHS group, meeting about cooperation with the commercial mail services. Time: Thursday October 3rd, 0920-1130 Place: UNI-C, Vermundsgt. 5, Copenhagen, Denmark 9 persons present. Harald Alvestrand Harald.Alvestrand@delab.sintef.no Marko Kaittola Marko.Kaittola@funet.fi Klaus Hansen khan@diku.dk Erik Lawaetz erik.Lawaetz@uni-c.dk Manu Mahonen Manu.Mahonen@funet.fi Bengt Ackzell Stig Ourier Alf Hansen Per Anderson What can we give to the PTTs? What can they give us? Round table: Norway: (Harald): Interconnection agreement between UNINETT MHS and TELEMAX. Nobody pays anything. Finland: (Marko): 2 commercial ADMDs: Elisa, use X.25 only. Mailnet, using X.25 or tcp/ip or ISO IP. FUNET pays for the use of Elisa and Mailnet. (10.000 FIM for service, and the X.25 connection is free; will probably drop payment for Mailnet because of "free" TCP/IP connection) Denmark: (Klaus): ADMD teledk exist. There is something called ADMD Business Mail. Informal group formed to discuss interconnection. The R&D MHS Service is still based on old Ean addressing. Public X.400 providers are using X.25 only. Goal is full integration on a non-pay basis. Sweden: (Per): Swipnet providing commercial ip service and also E-mail service. Tipnet is providing similar services. Large companies are using Memo and there are some gateways to the RFC world. Telverket provides td400. 70 PRMDs are connected. Some discussions with SUNET on how to connect to the SMTP world. Sender keeps all should be the principle. Registration of ADMD and PRMD names: In Sweden a national osi body will be formed this year. Bengt is the chairman of SC18 in Sweden. In Norway Statens Teleforvaltning is serving as a preliminary registration body. Finland has no registration body. Denmark is investigating how to register. Klaus has been given the task to write a document describing how this could be done. Conformance testing: No formal testing is done at the moment. Issues in Commercial ADMD - R&D MHS Interconnection: - Money, who pays the other? - Address space management? (Mapping) - Relay traffic Possible solusions: - Sender keeps all, at least for a temporary period of time. - Either org.no or org.prmd.admd.no The other direction: - National authority PRMD/ADMD - UNINETT authority PRMD UNINETT. - Interconnection agreements. Long discussion on funding principles. Agreement: Sender keeps all is a good principle as a start. When traffic patterns are determined, other principles could be negotiated. The PTTs have to be aware of that some of the traffic originated in the R&D community destined for the public customers, are from distribution lists in the R&D community, and are actually serving the public users. The address management proposed above, is a valid approach. The question of relay traffic is a part of the "money" discussion. In general, relay traffic should be allowed, and one should find forms of agreements that will allow this. We must accept that this may not be applicable for every destination. Some coordination between ADMDs in the Nordic countries are taking place in the Nordiska R}det. Could it be useful to let the NORDUnet Mail Inspector participate in these coordination meetings, to express the NORDUnet view?