[Nut-upsdev] snmp-ups shutdowns

Jim Klimov jimklimov at cos.ru
Thu Feb 27 14:43:25 UTC 2014

On 2014-02-27 14:21, Charles Lepple wrote:
> On Feb 27, 2014, at 7:07 AM, jimklimov at cos.ru <mailto:jimklimov at cos.ru>
> wrote:
>> This apparently implies that, unlike some docs say, the snmp-ups
>> driver can send the shutdown signals (is not crippled by design)? ;)
> We're working on it:
> http://buildbot.networkupstools.org/~buildbot/cayman/docs/latest/docs/man/snmp-ups.html#_shutdown
> A quick check says that the implementation has been there since v2.6.4,
> and the documentation caught up after v2.7.1. Without a SNMP UPS to test
> against in-house, you can imagine that I am not inclined to arbitrarily
> change things there without user feedback :-)

Ok, I'll have them give it a try :)

Also, is there any command to cancel a pending shutdown, if any was
requested previously - so as to reset the doomsday timer (so that, per
my earlier example, if several blades are shutting down and asking the
UPS to kill itself in a few minutes, every next blade that completes
its shutdown to this step, would give a few more minutes of uptime
to those which are still processing their stop-scripts)?

>> Also, are those oids apc-private or commonly used/standard among snmp
>> upses?
> is the APC prefix - I don't think many UPSes use other
> vendors' prefixes.

hmmm.... right, thanks :)

> This OID is curiously commented out in drivers/apc-mib.c, and has been since the beginning of our revision control history. I don't see why we couldn't add it back in as an instant command.
> I wonder if there was some confusion with . (upsBasicControlConserveBattery), which is currently mapped to "shutdown.return"
>>>> # turn off UP gracefully (APC) (stays off even if power is restored)
>>>> #UPS_OID=". integer 3"
> This should map to the "shutdown.stayoff" instant command.

And this means that the current code in the repository would not work
as expected, without fixing the above discrepancies first? Right?

//Jim Klimov

More information about the Nut-upsdev mailing list