CURRENT_MEETING_REPORT_ Reported by Gene Hastings/PSC and Dan Long/BBN Minutes of the User Connectivity Working Group (UCP) Summary A presentation on the UCP work-in-progress was made at the Farnet meeting on Monday. Useful discussion ensued and continued in the ORAD meeting Tuesday. The consensus was that the UCP work is important and should be actively pursued so that initial implementations are in place in the next 6-12 months. Encouraged by good suggestions and the support of Farnet and ORAD, the UCP group met twice and made good progress on these three projects: 1. A NOC PhoneBook o Reviewed collection efforts to date o NEARnet database forms-based entry being used o 22 NOCs have registered so far o Suggested new fields and formats (now incorporated in database) o Will advertize to wider audience soon o Plan to distribute PhoneBook with a finger-based search tool o All searches will also return caution that info is for NOCs only 2. Standardized Network Status Reports o Developed a syntax for standard email-based outage reports o NOCs will generate these reports which will contain info about current, past, or planned outages o These reports will be sent to a mailing list which anyone can subscribe to o People can develop their own tools for parsing and providing interactive access to this information o Ideally, end-users and NOCs could use such tools to get more information about connectivity problems 3. Standardized Trouble Ticket Handoffs o Revised the UCP Trouble Ticket Tracking draft to allow Network Service Centers to limit who they are required to accept calls from (from last meeting) o Network Service Centers will hand off tickets to other Network 1 Service Centers but will stay in the loop with the reporting user o Developed a syntax for standard email-based trouble ticket handoffs between Network Service Centers o Several groups are interested in participating in trials of this system For more information, join the list: ucp-request@nic.near.net DETAILED NOTES of the 11/19 meeting (first of two) by Gene Hastings The Distributed Agenda (roughly) o Status o NSC Phonebook - NISI - Current - Future o Reducing Need for Tickets - Notification Schemes - Database of Network Status - New Working Group? o UCP Ticket Sharing - What Information to Exchange - Method/Format for Exchange - Implementations Dan Long (NEARnet) gave an overview of FARNET's interest in UCP topics. NSC Phonebook - as of the meeting, there were 18 entries in the pilot NSC Phonebook. Note was made of the parallel NISI effort to collect similar listings for NICs. Vikas Aggarwal (JvNC) recommended that contact information be included in DNS TEXT records. The present NSC Phonebook Database is in Informix. Dan Long volunteered to continue to maintain it and to deploy a finger query agent for it. Dale Johnson (Merit) offered a second Informix host for it if someone else would maintain the actual db. Vikas volunteered to produce a DNS entry template. Group consensus was an acknowledgement that these are interim efforts. Reducing the need for tickets - that is, reducing the need for calls from users which require the opening of tickets. This might be secured through: 2 o Notification Schemes o Database of Network Status o End-User diagnostic tools These are all in keeping with the idea that if users are better informed about the state of the world, and have easier means of learning for themselves the nature of difficulties, they will have less need to call and talk to a person. An example of this at a department or campus level might be a bulletin board which lists scheduled outages and includes explanations of what services will be affected, along with pointers for further inquiries and help files explaining the nature of some classes of failure A strawman proposal was made to distribute email with a standard format, initially based on the NEARnet trouble tickets. Discussion followed as to which problem this proposal was intended to solve. Uses for standard format mail include: ease of information extraction when read; ease of parsing for inclusion in a database or for triggering alarms; assurance of completeness of information in report; and the possibility of making many reports machine-generated. Desired fields were felt to include: o ASN# o Net# o Net Name o Host Address/NSAP o Host Name o Affected protocol or service o Start/End Date & Time o Responsible Person or NOC o Ticket Cross-reference o Last Update o Reporting NOC o Perspective/Scope o EXPLANATION o FURTHER EXPLANATION There is still confusion and some disagreement concerning what things are or aren't tickets. There was, and will continue to be, discussion on use, control and interpretation. For example, whether these messages should be intelligible to, and distributed to, end users. Michael Patton (MIT) observed that poorly formed information distributed to the public would generate more calls, not less. Ittai Hershman (ANS) reported that nsr is now carried in a PSI newsgroup, so the mechanism for end-users to see those messages is in place. Further work on mail format was deferred until the meeting of the 20th. Discussion returned to the NSC Phonebook. New fields to add to 3 listings: o Administrator to escalate to o Domain name of NOC o Bigger net # field (allow listing of multiple net numbers) o Cross references to other nets, centers o Bigger phone number fields (multiple numbers) o FAX # o Discussion of upper vs mixed case for org and net names. [there is a practical limitation of the pilot db, in that it will not fold case for searches.] Questions were raised as to what limitations should be placed on the distribution of this information, if published. Following objections to having internal operations numbers available to arbitrary end-users, Ittai Hershman proposed limiting the distribution of the information to NOCs & NSCs, with harshly-worded boilerplate against indiscriminate release. A quick hack to limit availability is to include an access string in the finger query, acting as a pseudo password; Instead of ``finger psinet-nsc@nic.near.net'', something like ``finger psinet-nsc-abqothl@nic.near.net''. DETAILED NOTES of the 11/20 meeting (second of two) by Dan Long The second meeting focussed on mechanisms for handoffs of tickets between NSC's. We agreed that a similar format to that described above should be used to allow handoffs to be generated and parsed either manually or automatically. The group brainstormed a list of fields that would be of interest: o Description of problem o Description of solution o Location (or Source/Destination) of problem: AS#, Net#, Host Address, Service Description, etc. o Problem Start/End Date/Time o Ticket Open/Close Date/Time o Ticket Number (made unique by prepending a unique NSC identifier) o NSC List (list of NSCs that have handled this problem) o Notifications (who should be kept informed about this problem?) o Contact Info (who should be worked with to resolve this problem?) o Notes: number, date/time, author, text The group agreed that this list of items will likely need to evolve but that we should be conservative in the addition of fields so that the syntax remains simple and that the burden on human operators is minimized. There was a fair amount of discussion about notifications and whether end-users should be notified about steps taken by NSCs other than the originating NSC. Organizations have different 4 policies about how much detail to reveal. The consensus was that the originating NSC can use the Notifications field to include the user (or not) as they see fit and that other NSCs working on the problem should honor the notifications field to report progress. In the original paper by Matt Mathis, the idea was for the entire ticket to be handed off to the appropriate NSC and for the new NSC to deal with the user. We agreed on a change whereby the originating NSC maintains the contact with the user and keeps its own ticket open on the problem. It may, as the document describes, hand the problem off to another NSC but that NSC must then report back when it is done to the originating NSC who, in turn, will obtain closure with the user. The handoff will be handled much as the original document describes. The general format of the mail message will be: To: trouble-ticket-handoff@destination (the specific address for any given NSC is in the NSC PhoneBook) Subject: ticket-number {handoff, update, close} note-number (note-number is 0 on 1st handoff, 1 on 1st note, ... N+1 on close) And in the body of the message: Fieldname: (contents) ... ... Note: 1 Date Time Author (note text) ... ... Several people volunteered to begin using these formats for status updates and ticket handoffs. Dan Long to publish detailed writeup of formats so people can get started. Attendees Vikas Aggarwal aggarwal@jvnc.net Thomas Bajzek twb+@andrew.cmu.edu Robert Blokzijl K13@nikhef.nl Jeff Erwin Susan Estrada estradas@sdsc.edu Farrell Gerbode farrell@rice.edu John Gong jgong@us.oracle.com Eugene Hastings hastings@psc.edu Ittai Hershman ittai@nis.ans.net Dale Johnson dsj@merit.edu James Jokl jaj@virginia.edu Dan Jordt danj@nwnet.net Walter Lazear lazear@gateway.mitre.org 5 Daniel Long long@nic.near.net Donald Morris morris@ucar.edu David O'Leary oleary@sura.net Michael Patton map@lcs.mit.edu Marsha Perrott mlp+@andrew.cmu.edu Joe Ragland jrr@concert.net Ron Roberts roberts@jessica.stanford.edu Tom Sandoski tom@concert.net Bernhard Stockman boss@sunet.se Carol Ward cward@spot.colorado.edu Cathy Wittbrodt cjw@nersc.gov Paul Zawanda zawanda@ncsa.uiuc.edu 6