Date: Wed, 12 Jul 2000 21:47:20 +0000 (GMT) From: Terry Lambert <tlambert@primenet.com> To: abial@webgiro.com (Andrzej Bialecki) Cc: phk@critter.freebsd.dk (Poul-Henning Kamp), tlambert@primenet.com (Terry Lambert), freebsd-arch@FreeBSD.ORG Subject: Re: Final call for review: Dynamic sysctls. Message-ID: <200007122147.OAA24077@usr02.primenet.com> In-Reply-To: <Pine.BSF.4.20.0007122000550.52853-100000@mx.webgiro.com> from "Andrzej Bialecki" at Jul 12, 2000 08:05:37 PM
next in thread | previous in thread | raw e-mail | index | archive | help
> > >> Please obtain an OID arc from IANA, and establish your > > >> hierarchy under an OID, with both dynamic and static subtrees, > > >> so that these controls can also be exported via SNMP, ACAP, > > >> LDAP, SLPv2, Saluatation, HP JetSend, JINI, LISA, and T-Spaces, > > >> as well as other externalization protocols. > > > > > >Sounds like a good idea. How do I go about it? > > > > I already have registered a vendor OID with IANA for FreeBSD. > > Good! Then what it is? He can't just give it out and let you use it without allocating a sub-arc to the specific purpose. > > Terry on the other hand doesn't realize what he is saying here > > so don't pay too much attention to him for now. > > I disagree with you here. I certainly see merit in having a tree > that can easily be exported preserving 1:1 oid numbering wihtout fear that > they will clash with someone else's. Yes. More and more people (including IBM, since they bought Whistle) are using FreeBSD for thin servers. It would really piss me off if I couldn't put a thin server from one vendor on the network with a thin server from another, and use the same schema. > OTOH, since most oids are created nowadays with OID_AUTO, the > exact numbers will differ for the same nodes depending on how and > when they were created. Indeed, when we added OID_AUTO we more or > less abandoned the idea of having constant oid numbers that go > with names. As it is now, more reliable method would be actually > to create external oids from names using some hashing schema... :^)) Or have the OIDs _be_ the names, and use user space mappings for the humans who intend to deal with them. This would make them invariant, for the most part, if sparse. The big problem with variance and hashing is that you have to make it invariant, or get the same result on the hash, in order to be able to externalize the data. LDAP can externalize subschema or other attributes at the same time; this would be more difficult to achieve with SNMP, which doesn't intrinsically understand the idea of an indirect reference, and where schema is, more often than not, considered to be invariant, and as something that gets loaded into a management console and never changes. Even were you to do the indirection for LDAP, you would end up with a schema with its own statically assigned OID that then would use data, wherein the synthetic OIDs would just be some extraneous bit of information. Mapping from an LDAP modify request to a sysctl entry change on a particular machine would be a bear. 8-(. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-arch" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200007122147.OAA24077>