Quoth Nicolas Williams on Thu, Jan 15, 2009 at 03:00:13PM -0600:
> I understand now that "delmanifest" is intended to delete services --
> not manifests -- that were delivered by auto-imported services.

It doesn't delete manifest files, but it does delete the repository's
knowledge of such a file, and therefore the services they delivered.  In
any case, I'll revise the text to clarify this.

> A few comments:
> 
> a) A force option to "delete" seems more appropriate (and certainly less
>    confusing, if only to me).

That would facilitate the wrong mental model.  When a user removes
a manifest from /var/svc/manifest and wants to propagate that change to
the repository, requiring "delete -f" for the services it delivered
requires more work, magnifies the opportunity for error, and requires
special logic in svccfg to handle dependents properly.  It's a lot
easier for everyone involved for the user to just tell SMF which file he
deleted, and have the facility take care of the rest.

> b) Will one still be able delete, with the "delete" sub-command,
>    manually created instances of services delivered by manifests?
> 
>    I would expect the answer to be "yes."

Yes.  Such instances will be treated as administrative customizations.

> c) A service deleted with delmanifest will be autoimported again unless
>    the manifest [or pkg delivering is] removed.  This is just an
>    observation.  I don't know if this is a problem or not, though I
>    suspect it's not.

That's right.  The reason delmanifest exists is for users to propagate
the deletion a file from /var/svc/manifest.  If he uses delmanifest but
doesn't delete the file, then he's pretty much asking for the manifest
to be re-imported.


David

Reply via email to