150 likes | 174 Views
Netconf Notifications. Sharon Chisholm Hector Trevino IETF 67 November 2006. Issues. The -03 version of the Notification draft was well-reviewed. Most of those comments were resolved and incorporated into the -04 version. Exceptions were listed
E N D
Netconf Notifications Sharon Chisholm Hector Trevino IETF 67 November 2006
Issues • The -03 version of the Notification draft was well-reviewed. Most of those comments were resolved and incorporated into the -04 version. Exceptions were listed • https://ops.ietf.org/lists/netconf/netconf.2006/msg01223.html • This package includes unresolved issues from the last update and new issues raised against -04 on the mailing list. • They have been reordered to try to put the issues requiring more discussion earlier in the package
Issues – Data Models • Why do we need to query the notification subscriptions? [BL] • http://ops.ietf.org/lists/netconf/netconf.2006/msg01226.html • Does the schema contain the right information to query [AB]
Transport Mappings • Which transports must be supported in version 1.0? • - Document packaging • Include in Notifications RFC or create 2 separate RFCs? 4 separate RFCs? • - SOAP needs a lot of specification work; not going to hold up the entire • work item waiting for this document. [AB] • - SOAP could use Ajax Push design [TG] • http://ops.ietf.org/lists/netconf/netconf.2006/msg01216.html
Instance Validation • Changes introduced in later versions of the Netconf base protocol mean that instance documents of notifications (and rpc-replies), don’t appear to be validating • <xs:complexType name="dataInlineType"> • <xs:complexContent> • <xs:restriction base="xs:anyType"/> • </xs:complexContent> • </xs:complexType>
Issues – RPC Methods • create-subscription • why isn't there a stop-time in the replay capability? [BL] • why stop notifications after replay (e.g. no tail -f mode) [BL]
minAccess/ config-notconfig • Issue 9 from pre-03 list • The documentation currently defines appinfo to annotate information in the document. Primarily the maxAccess clause and some module Identity. We can’t publish with a dependency on this document since it would block us. • a. We talked about just publishing a short little document that would define the appInfo bits we needed, as appose to full data specification • b. Andy has suggested instead of maxAccess, moving to config or not config, which I think could also work. State versus statistics would also be a nice distinction to be able to make, but perhaps that can wait.
Relationships in Schema • The data model currently uses keyref to associate subscriptions to specific instances of named profiles • Is this too complex and under-described in document [AB]
Cardinality of Streams • Are you suppose to be able to associate more then one stream with a single subscription [AB]
Issues -Security Considerations • Need paragraph stating that security threats are handled during NETCONF session establishment, and the notification mechanism does not introduce any new threats, since any data available through <notification> can also be made available to the <get> RPC. [DR]
Misc Detailed Comments • Not necessary to allow combining named profiles and filters [AB] • stop notifications (2.3) • Wants to remove "or some mechanism outside the scope of this spec" • There is only <kill-session> unless and until the WG creates something else. [AB] • capability strings in 3.1 are wrong: • (correct format) • urn:ietf:params:netconf:capability:{name}:1.0
Issues - Syntax • General XSD issues) • Why do we need 3 namespaces? [BL] • The XML Schema should start with a top level element (e.g. <top>) and specify the containing elements all the way down. [BL] • http://ops.ietf.org/lists/netconf/netconf.2006/msg01226.html
Misc Comments • NETCONF stream • The default stream of NETCONF is well described. Doesn’t want it to be optional [AB] • Note in Montreal we said when it wasn’t specified it would be NETCONF • The schema does not reflect that it is optional.
Misc Comments • notification replay should not be a separate capability [AB] • Note that we agreed to a separate capability in Montreal • replayCompleteNotification • Why do we need to signal the end of replay with a special notification? Why is this detail needed at all? [AB]
Issue- Documentation • assorted comments and corrections (10/24/06 email) [BL] • http://ops.ietf.org/lists/netconf/netconf.2006/msg01227.html • assorted comments and corrections (10/24/06 email) [MB] • http://ops.ietf.org/lists/netconf/netconf.2006/msg01228.html