|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Status of DLB's technical commentsBased on list discussion and what's now on Julian's web site, here's what I think the status of my technical comments are. Most of these are resolved, and many thanks to Julian for his patient slog through these. At a high level, I think T.6, T.15, T.38 and T.39 are still open (see recent separate email on T.38 and T.39). I think T.9, T.12, and T.30 are resolved, but the resolutions require technical changes to what's currently in the working draft. All the rest are resolved, in some cases with editorial changes requested in the following list: T.1 - T.3. Fixes are ok with me. T.4 (2.2.6.3.1 .iqn - needs date unammbiguous) - Almost fixed. In the following text: This date MUST be a date during which the naming authority owned the domain name used in this format, and SHOULD be the date on which the domain name was acquired by this naming authority. the requirement that the domain naming authority owned the domain name at 00:01 GMT needs to be subject to the MUST, so the new text would read: This date MUST be a month in which domain name used in this format was owned by the naming authority at 00:01 GMT on the first day of the month, and SHOULD be the first month for which this was the case. T.5 (Synch and Steering bloated - for what we choose) New text is fine. I sympathize with Julian's comment on the wasted effort here - *please* save that text, as it may be useful elsewhere (e.g., RDDP WG architecture document). T.6 (Clarify text in 2.3 about discovery) I'm satisfied with the change of MAY to MUST with a list of the PDUs that are allowed on a discovery session, but whether to allow PDUs other than Send Targets and Logout on discovery sessions is an open issue. I don't care about this issue - I just want to see the MUST with a small list of what's allowed. T.7 (Portal defined by IP address does work through NAPT) No text change appears to have been made. I agree that this is a model issue that does not affect protocol operation, but use of an IP address to identify a network portal does need to take account of NAPTs. How about adding a statement that the IP addresses and TCP ports used to identify Network Portals are only assured to be valid identifiers within the containing Network Entity. They may not be valid or unique elsewhere due to the possible presence of Network Address Translators, Network Address Port Translators, private networks, etc. T.8 - Fix is ok with me. T.9 (SCSI port name inapropriate) No change has been made to the working draft. This should follow the agreement on the list to make the SCSI port name a string via conversion of ISID and TSIH to hex ASCII. Not fixed, but the fix appears to have been agreed to on the list. I agree with the comments on the list that the maximum string length of a SCSI port name should be 255 bytes. T.10, T.11 - Fixes are ok with me. T.12 (large-numerical-value does not cover lower than 2**64) List appears to have agreed that defining this in terms of allowed range rather than the actual value transmitted is the right thing to do. I agree. T.13 (clarify when 64k has to be supported) Text fix isn't quite right, as in using "e.g.", it fails to define "very long authentication items". It needs to explicitly say that the only authentication methods that use very long authentication are the SPKM methods in Section 10.3, so the 64k limit applies only when SPKM methods are offered or accepted on a connection. T.14 - Fixes are ok with me. T.15 (Make TPGT return on login MUST always) Changes made, but still open on the list, new working text contains some typos. T.16 (Second connection issue) Section 4.3.4 change to use SHOULD is fine. Swapping order of Sections 4 and 5 should remove the confusion about the text in the current Section 5. T.17, T.18 - Fixes are ok with me. T.19 (Conservative reuse in 8.1.1 SHOULD is appropriate) SHOULD is fine with me unless someone else wants to argue for MUST. T.20, T.21 - Fixes are ok with me. T.22 (Padding replace SHOULD with MUST be sent as 0) No change made - ok with me, consider this comment withdrawn. T.23 and T.24 - Fixes are ok with me. T.25 (Task reassign may need LUN) No change made. Not adding the LUN is ok with me. T.26 (Add Task Reassign to list of responses) Fix contains a typo, and the resulting list now contains all task management functions and hence could be replaced by a statement that the response is always sent. T.27 - Fix is ok with me. T.28 (Concern about discarding) Discussed on list, issue turned out to be editorial. Picking up latest text based on Mallikarjun's modification of my proposed new text will resolve this. T.29 - Fix is ok with me. T.30 (Resegmenting may come as a surprize - suggested a different request) Ouch - the new SNACK type code was inserted in the middle of the sequence making this change not backwards compatible with earlier versions. Please use 3 for the new R-Data SNACK and leave 0, 1 and 2 unchanged from -14. I am ok with the Initiator deciding whether it needs a new Response and using the existing Status SNACK mechanism to get it - that's definitely cleaner than my proposed abuse of the service response code. The current working draft uses a MAY for the initiator requesting a status retransmit - I prefer Mallikarjun's proposal on the list that the initiator MUST discard the first Response PDU and MUST use a Status SNACK to get one that is certain to reflect any resegmentation. I still disagree with Mallikarjun about the new R-Data SNACK type code - I would prefer to see this code used so that the initiator is clearly aware that it is in a situation where it MUST request status retransmission. Getting this wrong risks returning incomplete data on a READ. T.31 (Operational Error Level instead of supported) Fixed text is ok with me. There may be an open issue on the list concerning whether some sort of "ErrorRecoveryLevel 0.5" is permitted. T.32 - T.36 - Fixed text is ok with me. T.37 (Unsolicited data inclarity) Julian says "Already spec'd in 2.2.4", and it is. Please add references to Section 2.2.4 to the descriptions of the InitialR2T and BidiInitialR2T keys. This will be particularly effective if Section 2.2.4 is split into subsections as Brian Forbes has suggested. T.38 (IANA Text) Fixed text is ok, but will cause port 3260 to be abandoned by iSCSI for a port to be assigned by IANA at some indefinite time in the future. As I said in the comment, I think sticking with 3260 is better. I'll send a separate email to raise this issue on the list. T.39 (IANA registry for keys) Yes, this or something is needed, otherwise two vendors might assign different meanings to the value CRC64K resulting in a nasty interoperability failure. I just realized that we have a key namespace usage issue and possible key value namespace usage issues. Separate post coming on this topic. Thanks, --David --------------------------------------------------- David L. Black, Senior Technologist EMC Corporation, 42 South St., Hopkinton, MA 01748 +1 (508) 249-6449 FAX: +1 (508) 497-8018 black_david@emc.com Mobile: +1 (978) 394-7754 ---------------------------------------------------
Home Last updated: Thu Jul 11 14:19:01 2002 11272 messages in chronological order |