TUESDAY-------- DISMAN WG Meeting minutes Date : 07/16/2002 Time : 1545-1645 Room Number : 304 Reported by : Niraj Gopal (niraj@cisco.com) Chair Presentation: The chair mentioned that INTAP/OSMIC (http://www.intap.or.jp) is working on CIM/XML over SOAP for Alarms and that might be relevant for us. But no specific action is needed. XMLCONF might have some impact on us e.g. Juniper scheme does not send events thru XML. Condition MIB Prsentation: Action Item - Bert to resolve Boiler plate references for rfc1155, 1215, 1157, 2571 to informative/normative. Especially becasue SNMPv1 status is now historic and should be informative. Action Item - For security section, Kam to list the objects that are read-create or read-write and if they contain any privacy information. There was a discussion regarding referential integrity problem in one of the tables in this MIB. That MIB table uses a fully qualified Object Identifier of a resource (MIB Varable in another table) as one of its indices. But the value of that resource ID can change across reboots thereby causing this MIB table index to point to some other value, if the table is persisted. In addition, VACM configuration for such resource IDs can be problematic. It was asked if it would be adequate to say in the Security Considerations section that resources with unstable indices (non-persistent) can have problems. Bert pointed out that we need a separate doc as this a generic problem. Randy pointed out that ADSL profiles handle it thru description and that it is a known limitation of VACM. Action Item - A BCP might be needed for this general referential integrity problem. Shall we add a storage type to this table or save all entries? David Perkins pointed out that for CLI you cannot say if a particular instance be saved while others are not but it could be SNMP specific. Bert mentioned that either storage type or all entries saved is fine but it should be the same as we have been doing. Action Item - Kam to raise this question on mailing list if all entries should be persistent or only some. Action Item - Kam to define a IANAItuProbableCauseOrZero TC in this MIB with Unsigned32 syntax that can take any value of IANAItuProbableCause or 0. IANA should also be informed not to assign 0, in future, for IANAItuProbableCause. Bert pointed out that it could have a 0 value and 0 in Index is not good. Randy suggested that it is ok for this case. Rest of the presentation to be finished on Thursday. THURSDAY-------- DISMAN WG Meeting minutes Date : 07/16/2002 Time : 1545-1645 Room Number : 304 Reported by : Niraj Gopal (niraj@cisco.com) Chair Presentation: The chair mentioned that INTAP/OSMIC (http://www.intap.or.jp) is working on CIM/XML over SOAP for Alarms and that might be relevant for us. But no specific action is needed. XMLCONF might have some impact on us e.g. Juniper scheme does not send events thru XML. Condition MIB Prsentation: Action Item - Bert to resolve Boiler plate references for rfc1155, 1215, 1157, 2571 to informative/normative. Especially becasue SNMPv1 status may be going historic and should be informative. Action Item - For security section, Kam to list the objects that are read-create or read-write and if they contain any privacy information. There was a discussion regarding referential integrity problem in one of the tables in this MIB. That MIB table uses a fully qualified Object Identifier of a resource (MIB Varable in another table) as one of its indices. But the value of that resource ID can change across reboots thereby causing this MIB table index to point to some other value, if the table is persisted. In addition, VACM configuration for such resource IDs can be problematic. It was asked if it would be adequate to say in the Security Considerations section that resources with unstable indices (non-persistent) can have problems. Bert pointed out that we need a separate doc as this a generic problem. Randy pointed out that ADSL profiles handle it thru description and that it is a known limitation of VACM. Action Item - A BCP might be needed for this general referential integrity problem. Shall we add a storage type to this table or save all entries? David Perkins pointed out that for CLI you cannot say if a particular instance be saved while others are not but it could be SNMP specific. Bert mentioned that either storage type or all entries saved is fine but it should be the same as we have been doing. Action Item - Kam to raise this question on mailing list if all entries should be persistent or only some. Action Item - Kam to define a IANAItuProbableCauseOrZero TC in this MIB with Unsigned32 syntax that can take any value of IANAItuProbableCause or 0. IANA should also be informed not to assign 0, in future, for IANAItuProbableCause. Bert pointed out that it could have a 0 value and 0 in Index is not good. Randy suggested that it is ok for this case. Rest of the presentation to be finished on Thursday. Disman Thursday 3:30 18 July 2002 Reported by Chris Elliott chelliot@cisco.com Alarm MIB Alarm 38: This issue needs to be clarified on the mailing list. Alarm 42: Sharon gets to do what she wants with this and will post her decision for comments on the mailing list. Alarm n+2: Reserve 0 in TC and point IANA to this TC so they know to not allocate 0. Clustering and Replication No consensus on if this would be useful. Needs further discussion on mailing list. MIB table modification tracking MIB Meeting seemed to have a fair amount of interest in developing this MIB. Should it include all objects in table, exclude all counters, or exclude all objects by default? Needs further discussion on mailing list Distributed Management Apps. No consensus whether the IETF is the appropriate org for this or whether SNMP has a role here. No action required. Continued discussion should be on ops-nm mailing list. Charter Updates Have two target dates overdue? Alarm MIB will be updated and go WG last call August. ARC MIB update next week and will go WG last call then. Can disman go dormant? Consensus of meeting (not list) was that there is still interesting things to work on.