During the ISE review of the proposed RFC, the IETF editor has asked for clarification of the LOGIN command, since he, like most people, assumes that "LOGIN" means something like "login to a shell".

The Developer Guide Chapter 9 https://networkupstools.org/docs/developer-guide.chunked/ar01s09.html "Network Protocol Information" says

 9.9. LOGIN

 LOGIN <upsname>    Response: OK (upon success) or various errors

 Note This requires "upsmon slave" or "upsmon master" in upsd.users

 Use this to log the fact that a system is drawing power from this UPS. The
 upsmon master will wait until the count of attached systems reaches 1 - itself.
 This allows the slaves to shut down first.

 Note You probably shouldn’t send this command unless you are upsmon, or a
 upsmon replacement.

 9.4 GET NUMLOGINS <upsname>
 Response: NUMLOGINS <UPSNAME> <VALUE>

 where <value> is the number of clients which have done LOGIN for this UPS. This
 is used by the master upsmon to determine how many clients are still connected
 when starting the shutdown process.

I propose saying:

Figure 4                                 "The client"
           ,--------------------,---------------------,
 ,-----,   |     UPS       <-Commands      Primary    |
 |     |---|  Attachment        |         Management  |   Primary
 |     |===|    Daemon       Responses->    Daemon    |
 |     |   '--------------------'---------------------'
 | UPS |            ^
 |     |            '<-Commands---Responses->,
 |     |                                     v
 |     |            ,--------------,-----------------,
 |     |============|              |     Secondary   |
 /-----\            |              |     Management  |   Secondary
                    |              |       Daemon    |
                    '--------------'-----------------'

LOGIN is not the conventional user access to a shell. In a configuration such as Figure 4 in which a UPS protects more than one system, the Primary (2.8) Management Daemon (2.6), upsmon, needs to know how many Secondaries (2.9) are currently "active", i.e. powered by the UPS, either on wall or battery power. The Attachment Daemon (2.1), upsd, supports this by keeping a count of all the "active" systems powered by a UPS. The count is initialised, one secondary at a time by the LOGIN command, which should be understood as "count this secondary as active". LOGIN is one of a trio of commands for Secondary (2.9) counting: command LOGOUT (4.2.7) decrements the count and a Management Daemon (2.6), upsmon, may read the count at any time using command NUMLOGINS (4.2.2.3).

The LOGIN command is also sent to the Attachment Daemon (2.1), upsd, for the Primary (2.8) so during normal, fully protected operation, the count is 1 (the Primary (2.8)) + the number of Secondaries (2.9). During a full system shutdown, the count drops as each Secondary (2.9) Management Daemon (2.6), upsmon, executes command LOGOUT (4.2.7) during its own shutdown. When the count drops to 1, only the Primary (2.8) is "active" and it knows that all the Secondaries (2.9) have shut down.

Is that correct?

Comment: had the command LOGIN been called SETACTIVE, with the upsmon flag ST_LOGIN changed to ST_ACTIVE, and NUMLOGINS changed to NUMACTIVE this mechanism would probably be easier to understand. LOGOUT might be NOTACTIVE.

  Current   Proposed
  LOGIN     SETACTIVE
  LOGOUT    NOTACTIVE
  NUMLOGINS NUMACTIVE
  ST_LOGIN  ST_ACTIVE

Roger
_______________________________________________
Nut-upsuser mailing list
[email protected]
https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/nut-upsuser

Reply via email to