Summary

  • RFC 1289 put DECnet Phase IV management objects into the Internet MIB so TCP/IP and DECnet systems could be managed through a common SMI/MIB/SNMP environment. Its mandatory, read-write state objects named ON, OFF, SHUT and RESTRICTED.
  • The labels described different operational conditions. SHUT barred new logical links while retaining existing ones until they disappeared; OFF additionally terminated existing links and stopped routing; RESTRICTED only barred new incoming logical links. A represented state therefore did not prove a completed shutdown, an absent route, authorization, reachability or any external result.

A common management surface did not make different systems one state machine

RFC 1289 arrived in December 1991 as a standards-track extension to the Internet MIB. It was written for an environment in which TCP/IP-based internets and DECnet Phase IV systems could coexist, and in which management stations using the Internet SMI, MIB and SNMP could work across that combination. The ambition was integration of management, not the erasure of the systems being managed.

That distinction appears in the document's object design. DECnet objects are placed into the Internet MIB, their names and syntax are adapted to standard MIB form, and an object can be exposed through a shared management interface. None of that makes the object a complete account of the machine beyond it. A state value is still a bounded description with defined transitions and effects. Its usefulness depends on retaining the boundaries in the definition rather than letting a familiar label carry every conclusion an observer wants.

The RFC even records an apparently small translation choice: the state integers are incremented by one from their standard DECnet values in order to comply with RFC 1155. The point is not numerical trivia. A management representation can change the coding of a state without changing what the state means in the network. A compatible integer is a way to carry a distinction; it is not proof that an action occurred, that an actor was entitled to take it, or that the action has finished.

Four values separated admission, drain, termination and routing

phivSystemState is mandatory and read-write, and its descriptions give the state names their limits. ON allows logical links. OFF allows no new links, terminates existing links, and stops routing traffic through the node. This is the strongest of the four descriptions, but it is still a statement about the defined operational state. It does not identify a particular node in the world, confirm a command transcript, or establish that every application depending on that node has observed the consequence.

SHUT is the value that prevents a shortcut. It allows no new logical links. It does not destroy existing logical links. It goes to OFF when all logical links are gone. The definition thus carries a condition between a decision to stop admitting new work and the later OFF state. Someone reading only the word “SHUT” loses the waiting surface: there may be existing logical links, and the RFC specifies that their disappearance is what completes the transition.

RESTRICTED is not merely a gentler spelling of shutdown. It allows no new incoming logical links from other nodes. The RFC does not give it the OFF description's termination of existing links or cessation of routing. Nor does it say that no outbound action, local work or current connection can exist. It denotes a narrower admission boundary. A system can become less open to a particular class of new request without becoming unavailable in every sense.

These are not three degrees on one vague dial. They partition consequences. Admission of new logical links, preservation of existing logical links, termination, routing traffic and the direction of a new connection are different surfaces. The value name matters only because the RFC says which surfaces it reaches.

SHUT named a transition whose completion lay elsewhere

The value is especially useful as a historical lesson because it resists the false comfort of instantaneous state. In ordinary language, a shutdown sounds completed when it is announced. RFC 1289 makes it conditional. No new logical links may enter, but old links persist; the system reaches OFF after all are gone. That is a design for drain, not a synonym for immediate removal.

The difference changes what evidence can support. A SHUT value can support a claim about the management semantics defined by the RFC. It cannot by itself support “the node was off,” “all sessions had ended,” “routing had stopped,” or “the affected service was unavailable.” Each of those claims needs a later stage or another observation. The OFF description includes termination and stopped routing; the SHUT description does not say those things have already happened.

This is not pedantry about a legacy protocol. It is a method for reading operational records. A requested state, an admission rule, a draining workload, a routing change and a user-visible result are often arranged as if they were one fact because they happen near one another. RFC 1289 insists on separate language. The earlier stage may make the later one possible, but it does not inherit the later stage's evidence.

The remote entry repeats a state vocabulary, not a global observation

RFC 1289 also defines phivEndRemoteState, a mandatory, read-write entry representing the operational state of the remote node being evaluated. It repeats the same four values and their descriptions. That repetition is important but limited. It shows that the same state vocabulary was intended to apply at a remote-node entry. It does not give the table the power to prove a remote machine's current reachability, identity, authorization, physical condition or the completion of a management action outside the object semantics.

An entry can name a remote node and carry a state field without becoming a universally complete observer. The RFC's own wording keeps the entry tied to the remote node “being evaluated.” It does not turn a value into an independent witness for every route, service, user or administrative decision around the node. A management record remains a record at its defined surface.

Read-write was an access classification, not a retrospective mandate

The ACCESS read-write declaration matters because it tells implementers that the object is designed to be written as well as read. It does not answer who held credentials, which access controls existed, whether an attempted write was accepted, whether a device implemented the object, or whether a person had authority to impose a state on an operating system. Those questions sit outside the facts RFC 1289 supplies.

The temptation to turn a writeable field into a mandate is strong because it produces a simple narrative: an administrator wrote SHUT, so the node shut down. The RFC offers a more useful account. It supplies a vocabulary for defined states and a condition for moving from one of them to another. Any real execution still requires evidence of the command, the target, the applicable controls, the existing links and the observed completion.

Sources and evidence limits

This article uses RFC 1289 — DECnet Phase IV MIB Extensions. The source supports its December 1991 standards-track status, the common management context, the mandatory read-write state objects, the definitions of ON, OFF, SHUT and RESTRICTED, the remote-state entry and the RFC 1155 integer adjustment. It does not establish a current DECnet node, a submitted command, authorization, access-control configuration, drained links, stopped routing in a particular deployment, current reachability, a live service or any user-visible outcome.