 |
 |
 |
 |
|
 |
|
 |
|
 |
|
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
Hi
Su.
You're
right, we should usually consider prefix length before precendence, so it is
better to delete rule 3, not rule 5. I'll make that change when I
next update the MIB.
Yes,
pimStaticRPOverrideDynamic does override the decision procedure described by
pimGroupMappingTable. You're right that it I should mention it
there.
However, note that these rules are a gray area where we are providing an
outline of implementation-specific algorithms so that the MIB operator can
understand what is happening. We do not want to provide a completely firm
specification, because that is up to the implementation. See rule 6 for
example.
Finally about pimRPMappingChange, note that the origin and group prefix
are present in that notification, even though they are not explicitly
listed. They are not-accessible index fields, which are present whenever
objects indexed by them are present. So all the fields from a
pimGroupMappingEntry are available in a pimRPMappingChange
notification.
Thanks
for your comments,
David
McW.
-----Original Message-----
From: Su Haiyang [mailto:suhaiyang at huawei.com] Sent: 27 February 2007 12:46 To: David McWalter Subject: RE: [pim] Some doubts about draft-ietf-pim-mib-v2-09? Hi, DmcW, Thank you for your reply. There are still other doubts in my mind, please refer to the following comments with blue fonts: From: David McWalter
[mailto:DMcW at dataconnection.com]
Hi Su.
1. What's the purpose of pimGroupMappingPrecedence item? Is it only used in mapping a group prefix to a static rp? [DMcW] If more than one RP could be used for a group, the pimGroupMappingPrecedence chooses between them.
2. If there are static RP (embedded RP added in IPv6 multicast), BSR-RP, or some other dynamic RP mechanism like CISCO’s private AutoRP, willing to sever a group/prefix, How to choose the unique RP? [DMcW] Each willing RP has a pimGroupMappingEntry row. The one with the lowest numerical pimGroupMappingPrecedence is chosen.
[Su] Shouldn’t the pimGroupMappingEntry row with greatest value of pimGroupMappingGrpPrefixLength be chosed first，and then the row with lowest numberical pimGroupMappingPrecedence？So the following rule 4 should be considered before rule 3 if rule 5 has no effect.
Another doubt, there is a pimStaticRPOverrideDynamic item defined at PimStaticRPEntry of pimStaticRPTable, and the following is commented at pimStaticRPPrecedence’s description, “If pimStaticRPOverrideDynamic is set to TRUE, all dynamic RP configuration is overridden by this static configuration, whatever the value of this object.” My doubt is, why RP mapping rules defined at pimGroupMappingEntry dosen’t consider whether static RP is preferred to dynamic RP mechanism?
Another doubt about Page 70 of draft-ietf-pim-mib-v2-09, pimRPMappingChange trap is defined as bellow
pimRPMappingChange NOTIFICATION-TYPE OBJECTS { pimGroupMappingPimMode, pimGroupMappingPrecedence } STATUS current DESCRIPTION "A pimRPMappingChange trap signifies a change to the active RP mapping on this device.
This notification is generated whenever the counter pimRPMappingChangeCount is incremented, subject to the rate limit specified by pimRPMappingChangeTrapPeriod." ::= { pimTraps 4 }
Shouldn’t origin and group prefix be added into above trap so as to know which mapped group prefix changes?
3. Why the pimGroupMappingPrecedence appear twice in the above words(row3 and row5)? Is there any diffrence between this two rows? [DMcW] You're right, 3 and 5 say the same thing. I think 5 has no effect, and should be deleted.
Hope that helps, David McWalter.
-----Original
Message----- Hi, I have some doubts about The PIM Group Mapping Table defined at draft draft-ietf-pim-mib-v2-09 I will appreciate your help if someone can give answer about the following doubts in page 66 of draft-ietf-pim-mib-v2-09, there are some words as bellow:
Given the collection of rows in this table at any point in time, the PIM mode and RP address to use for a particular group is determined using the following algorithm.
1. From the set of all rows, the subset whose group prefix contains the group in question are selected.
2. If there are no such rows, the behavior is undefined.
3. If there is at least one such row, from the selected subset of rows, the subset that have the lowest value of pimGroupMappingPrecedence are selected.
4. From the selected subset of rows, the subset that have the greatest value of pimGroupMappingGrpPrefixLength are selected.
5. If there are multiple selected rows, the subset that have the highest precedence (the lowest numerical value for pimGroupMappingPrecedence) are selected.
6. If there are multiple selected rows, the row selected is implementation dependent; the implementation might or might not apply the PIM hash function to select the row.
7. The group mode to use is given by the value of pimGroupMappingPimMode from the single selected row; the RP to use is given by the value of pimGroupMappingRPAddress, unless pimGroupMappingOrigin is 'embedded', in which case the RP is extracted from the group address in question."
I don’t understand these words totally: 1. What's the purpose of pimGroupMappingPrecedence item? Is it only used in mapping a group prefix to a static rp? 2. If there are static RP (embedded RP added in IPv6 multicast), BSR-RP, or some other dynamic RP mechanism like CISCO’s private AutoRP, willing to sever a group/prefix, How to choose the unique RP? 3. Why the pimGroupMappingPrecedence appear twice in the above words(row3 and row5)? Is there any diffrence between this two rows? Best Regards Steven Su
|
_______________________________________________ pim mailing list pim at ietf.org https://www1.ietf.org/mailman/listinfo/pim