Minutes of the 42nd IETF meeting Chicago, August 1998 Bridge Working Group Minutes recorded by Les Bell RFC 1525 Keith McCloghrie gave an overview of the status of RFC 1525 and a proposal for how to move it to a full standard. Response from the mailing list indicated that Cisco and IBM have implemented this MIB, though not fully. It is proposed to deprecate those items that have not been implemented, and to optionally allow all read-write items to be implemented as read-only. It is not proposed to re-write this MIB in SMIv2 syntax. Bridge-MIB Extensions Changes in ietf-bridge-bridgemib-draft-01.txt The changes since the previous draft were reviewed, with the opportunity for the group to comment on the issues before they are closed. The group accepted the changes in the current draft with the following comments: Issue 31 - VLAN Bridge up-time - it was noted that it would be easy to implement this per- VLAN, but the group decided not to change the current draft. Issue 33 - The compliance clauses for the service requirements objects are to be defined as an optional group, instead of with "MIN-ACCESS not-accessible". Bridge-MIB Extensions Open Issues in ietf-bridge-bridgemib-draft-01.txt - The open issues without proposed resolutions in this draft were discussed, with recommendations (from Andrew) for closing them. Issue 5 - Use of the ifStackTable - a proposal on how this should be used, complementing the Q-BRIDGE-MIB, should be made to the mailing list. Feedback from this proposal should be added to section 3.4.2.2 of the next draft. Issue 50 - The range restrictions should be removed from dot1qVlanFdbId, dot1qMaxSupportedVlans and dot1qNumVlans. Issue 51 - dot1qConfigurablePvidTagging - the description of this object should be modified to indicate that attempts to set more than one VLAN to be untagged on egress may be rejected by devices which do not support this option. Issue 55 - dot1qStaticUnicastAllowedToGoTo should not be split into 'Fixed' and 'Forbidden' port lists. No change is required. Issue 56 - The value of '0' for dot1qTpFdbPort is required to indicate an address may defined in the static table, but has not been learnt on any port yet. No change is required. Issue 57 - Change dot1qVlanStaticUntagged to dot1qVlanStaticUntaggedPorts as proposed. Issue 58 - The value of dot1qVlanStaticName does not need to be unique, as proposed. No change is required. Issue 59 - dot1qPvid should have syntax 'VlanIndex' (instead of 'VlanId') as proposed. Issue 60 - Clarify the descriptions of dot1qPortAcceptableFrameTypes and dot1qPortIngressFiltering to indicate they also apply to GMRP packets. Issue 61 - dot1qLearningContraintsLastChange should be removed. Issue 62 (not in the current draft) - The proposal to remove support for multiple egress ports in static unicast FDB entries is not applicable, as they indicate ports the address MAY be learnt on, not ports they MUST be forwarded to. No change is required. Issue 44 - It was agreed that {per-port, per-VLAN} counters are useful, but the full set defined by 802.1Q is not required. A subset similar to the per-port counters defined by RFC 1493 should be adopted: dot1qTpVlanPortInFrames, dot1qTpVlanPortOutFrames and dot1qTpVlanPortInDiscards. 'InDiscards' does not include error frames. 'DiscardOnIngressFiltering' is not counted. 'DelayExceeded' is only counted per-port (RFC1493). Optional 64-bit and 32-bit versions are to be defined. These counters apply to transparent and SRT bridges Issue 49 - A new object should be added for the 'next free vlan index' available for local vlans without a standard 802.1Q vlan tag. The value of this object may be read by a manager and used to create a new local vlan entry. This value will automatically change when the current value is used. Issue 52 - An explicit "dot1dExtendedFilteringServicesStatus" is not required to enable/disable this feature. This can already be achieved by disabling GMRP and providing appropriate entries in the dot1qForwardAllTable. No change is required. Bridge-MIB Extensions New Issues with ietf-bridge-bridgemib-draft-01.txt David Melman raised the following issues, with the promise of more to come on the mailing list. Issue 72 (not in the current draft) - The default value for 'dot1qForwardAllStaticPorts' is defined as 'a string of ones of appropriate length'. The description needs to be clarified to indicate this applies only to ports in the vlan for each entry. This applies to several of the PortList objects in the MIB. Issue 76 (not in the current draft) - The statistics counted per-port/per-VLAN can be excessive, so it was suggested that a control table could be defined, in which the manager could define a sub-set of the port/VLAN combinations for statistics to be collected. The group did not seem to think this was worthwhile. RFC 1493 - Keith McCloghrie will poll for implementation experience on the mailing-list, and many folks volunteered to respond.