Re: RE: V2 Changes for next RFI v2 MIB (draft 10) - Clarification

Hi -

> From: "Eduardo Cardona" <[email protected]>
> To: "Minnie Lu" <[email protected]>; "Randy Presuhn" <[email protected]>
...
> Sent: Friday, April 16, 2004 11:58 AM
> Subject: RE: [ipcdn] RE: V2 Changes for next RFI v2 MIB (draft 10) - Clarification
...
> The only pending part.
>   "For non-active row, the value is always 'volatile(2)'"

This seems like an unusual constraint.  My understanding is that
the point of having a writable StorageType in a table with RowStatus
is to permit the management application to specify the desired
persistence behaviour of the row.  Regardless of the specified
storage type of a row that is under creation, if it stays in a non-active
state too long, the agent needs to delete it.  (Think of it as preventing
a memory leak in object identifier space.  :-)

> I am not sure if the active to non-active Rowstatus transitions for
> StorageType are needed or what are the current practices, I won't think
> it hurts, will like to hear Randy's and other's opinions
...

The most important active to non-active transition is probably "destroy",
which is necessary for obvious reasons.  Explicitly placing a row
"not in service" is useful if there is a need to modify it, and changes
to the active row would not be permitted for some reason.

The important thing is to be clear on how you want it to work.
I think part of the difficulty here is that it's beginning to sound
like the lifecycle of these rows created for cloning may be quite
different from other rows.  For example, "destroy" doesn't make
sense for "normal" rows here.  Since provisioning isn't supported,
"create-and-go" and "create-and-wait" also don't make sense for
"normal" rows.

As I indicated before, the table is being used for two very
different purposes, and, if it weren't so late in the discussion,
I'd advocate a different approach.

As to Minnie's point, it'd be perfectly reasonable from a MIB design
perspective to not use the StorageType TC, and in the table
description to require active rows to persist.  The question is
whether you actually want it to work that way.  (Forgive me if I've
misunderstood Minnie's concerns.)  What StorageType brings is:
   - a way to control whether entries created by management persist
   - a way to learn whether an entry will persist
   - a distinction between nonVolatile, permanent, and readOnly entries.
If these features aren't needed, then there's a legitimate question
whether it's the right tool for the job.

Randy