Editor's Note: Minutes received 12/20 CURRENT_MEETING_REPORT_ Reported by Steve Waldbusser/CMU Minutes of the Host Resources MIB Working Group (HOSTMIB) Agenda o Administration o Identify New Issues o Resolve Issues - Queues - Networked printers - In hrPrinterPrintPages, what is a page? o Plan/Conclusion New Issues o swRunCPU o Parent ID o User ID o Terminal ID o International character set o Remote file systems (size, use, etc.) o Chassis MIB (as it relates to HostMIB) o Processor speed o Remote FS Discussion and Resolution of Issues 1. Queues (Three Choices Proposed): o Don't do queues. o Overload hrStorage to include queues. o Add a new table. The question was posed as to whether the host resources MIB was the appropriate place for queue management? One point in favor was that queues are a good metric of system performance. Against, queues are very specific to their use. In other words, a print queue is different from a mail queue. We can't do a decent job of representing queue specifics here in the host MIB. In conclusion the Group unanimously agreed that we should not manage queues in the host MIB. 1 2. Network Printers Currently printer support is for local printers. Remote FS is good. Remote printer could be useful but is very different. PRO: Connecting to remote FS is the same (from a users perspective) as connecting to a remote printer, especially on a PC. CON: Remote FS is knowledge that is integral to the host OS. Printing is stateless and unknown by the host OS. Can we do a reasonable job of monitoring a network printer. Is this a PC specific problem? Should networked printers be handled in a printer MIB? Conclusion: Do not support networked printers as a part of this MIB effort. This discussion uncovered that we should add wording to MIB to indicate that we show all FS's that are system wide resources (as opposed to FS's which are specific to a user). 3. What is a Page? Proposal Number One is to drop hrPrinterPagesPrinted, and count octets instead. An observation was made that 80% to 90% of the printers can not count octets. Proposal Number Two is to drop pagecount and add ``printer needs service'' bit in printer status. This bit is an OR of the other need service bits AND any other service the printer knows it needs. Conclusion: The Group agreed to support proposal Number Two and add the definition of BIT STRINGs from the snmpV2 effort to this MIB, encoded as an OCTET STRING. 4. swRunPerfCPU Conclusion: A proposal was made to change this to count centi-seconds run rather than percentage. The advantage to this scheme is that agent can look to process table (in typical OS environment) rather than have to maintain count at each interval (or task switch). In multiprocessor environments, this count is the sum of this process across all processors. (Parent Process ID, user ID, terminal ID) Conclusion: These are very OS specific and not appropriate in this MIB effort. 5. International Character Sets Discussion: Should we tackle this complex problem in this MIB effort? Is NLS support an SNMP problem and outside of our scope? Proposal: Create a textual convention (TC) called ``characters'' which is an octet string. NMS should parse this to decide if it is printable ascii. Conclusion: Modify the MIB to add the characters TC. 6. Size of Remote FS 2 Conclusion: This is OS specific, so the Group agreed to withdraw this. 7. Chassis MIB relationship to host MIB Discussion: The chassisMIB is facing some interesting similarities to the hostMIB. Conclusion: Interested individuals can work with the chassisMIB Working Group. However, we won't do this as a function of the HOSTMIB Working Group. 8. Processor Speed Conclusion: Delete this as it is covered by ProductID. Planning/Conclusion Where do we go from here? The Group reached consensus that all appropriate substantive issues had been addressed and that the MIB should be submitted to the IESG for publication as an RFC at the Proposed Standard level. In particular, this consensus was that the appropriate amount of effort had been placed in identifying issues, and that each of these issues had been resolved. Any further issues that might be identified after the meeting would be addressed when the MIB is to go to Draft Standard. Bug fixes and clarifications during the final editing process would be agreed upon on the mailing list. The plan for the document is that Grillo and Waldbusser will make changes based on the results of the meeting and make that document available via FTP and a notification on the mailing list. A week or two of discussion would occur on issues of bug fixes and clarifications. This would result in editing instructions that would be applied to the MIB and the MIB would then be submitted to the internet-drafts directory and to the IESG. Attendees Bill Bowman bill@hpprsd.mayfield.hp.com Stephen Bush sfb@ncoast.org Chris Chiotasso chris@andr.ub.com James Davin davin@bellcore.com Michael Davison davison@fibercom.com Manuel Diaz diaz@davidsys.com Jon Dreyer Jon.Dreyer@east.sun.com Raymond Edgerton bellmf.edgerton@uwm.edu Daniel Fauvarque dfauvarq@france.sun.com Pete Grillo pl0143@mail.psi.net Gerd Holzhauer holzhauer1@applelink.apple.com Kevin Jackson kmj@concord.com Ronald Jacoby rj@sgi.com 3 Mark Kepke mak@cnd.hp.com Zbigniew Kielczewski zbig@eicon.qc.ca Dwain Kinghorn microsoft!dwaink@uunet.uu.net Andrew Knutsen andrewk@sco.com David Lin lind@janus-ccm.zenith.com William McKenzie mckenzie@ralvma.vnet.ibm.com Patrick Mullaney mullaney@ctron.com David Orelowitz david@c1st.microcom.com Gaige Paulsen gaige@intercon.com Ed Reeder ereeder@vnet.ibm.com Bradley Rhoades bdrhoades@mail.mmmg.com Dan Romascanu dan@lannet.com Assaf Rubissa asaf@fibhaifa.com Jon Saperia saperia@lkg.dec.com Michael Sapich sapich@conware.de Timon Sloane timon@rahul.net Dean Throop throop@dg-rtp.dg.com Panos-Gavriil Tsigaridas Tsigaridas@fokus.berlin.gmd.dbp.de Steven Waldbusser waldbusser@andrew.cmu.edu 4