clarification text for docsIfCmtsModulationTable re: row persistency
"Jean-Francois Mule" <[email protected]>
| Newsgroups | gmane.ietf.ipcdn |
|---|---|
| Message-ID | <[email protected]> |
Andy wrote:
> The proposed additional sentence for docsIfCmtsModControl
> should be more explicit.
>
> Instead of, "In some implementations, rows created at run
> time may not be deleted", state "In some implementations,
> pre-defined rows created at system initialization time may
> not be deleted"
Based on a chat with Eduardo, I do understand a little better that the persistence of those rows may depend on the product implementation. However, I'm curious what this implementation-dependent behavior means to operators & sys admins?
I am a bit concerned that adding such vague statements don't help any of us in the end (they don't help sysadmins because one cannot assume much and they don't help new implementors because they don't give much guidance).
See below.
Minnie wrote:
> I would like to propose that only the those entries with
> active value in
> docsIfCmtsModControl object MUST be saved after the managed
> device restarts.
I like this proposal better because it states what the behavior should be for *all* implementations on a subset of entries ('active' rows only, right?). Are they any concerns with this approach?
Minnie wrote:
> For those notInService entries, it is vendor specific
> implementation to save them or not.
What about 'notReady'? To me 'notReady' and 'notInService' should fall in the same bucket.
Minnie wrote:
> Those entries which rowStatus is notInService might not have
> all consistent
> attributes and they are not ready to be used (that is why the
> rowstatus is
> notInService). I don't think there is a need to save those
> notInService
> rows. But some vendor might want to save them, so I propose
> to leave it to
> vendor's choice.
I personally think it is acceptable to not mandate the persistency of data for 'notReady' and 'notInService' and that we can justify it as you proposed. It may also convey Andy's comment about implementation specific behaviors - but it constraints those to a subset of rows for which nothing was certain from a mgmt station point of view.
How about the following text:
Entries with a status column (docsIfCmtsModControl) of value 'active' MUST be persist across reboots. Entries with a status column (docsIfCmtsModControl) of value 'notReady' and 'notInService' may not persist after reboots since those entries were explicit made not be available for use by the managed device ('notInService') or the agent had not sufficient info anyway ('notReady').
Comments?
Jean-François