[Nut-upsdev] blazer_usb Shutdown

[Charles Lepple]

(please use reply-all - this list does not alter reply-to headers)

On Sep 10, 2015, at 6:23 AM, Thomas Aichinger <taich at qpl.at> wrote:
> 
> Hi,
> 
> I am using a ALLNET 91500 UPS with blazer_usb driver
> 
> /usr/lib/ups/driver/blazer_usb -a ups91500 -L
> Network UPS Tools - Megatec/Q1 protocol USB driver 0.10 (2.7.1)
[...]
> I set delay.shutdown to 600 sec but UPS turns power off after about 120 sec.

The shutdown delay gets discretized into a few different values, and is then sent to the UPS:

http://www.networkupstools.org/protocols/megatec.html#_shutdown_command

I'm not too familiar with the internals of this driver (development has shifted to its successor, nutdrv_qx, which should also be available in v2.7.1) but it looks like you can manually test the shutdown with the following command (after everything else has been stopped; perhaps send this from another system?):

/usr/lib/ups/driver/blazer_usb -a ups91500 -DDD -k

You should see a debug line with "send: S10R000".

Although it appears that a 600-second delay is within spec, you could try 540 to see if it makes a difference. It is also possible that the UPS is hardwired for a two-minute delay.

> I also tried to raise FINALDELAY to 600 and changed SHUTDOWNCMD to shutdown -h +10

These don't affect the shutdown command that is sent to the UPS.

> This is quite useless ups always turns power off after about 120 sec, regardless how much charge is left.
> Since my ESX needs about 5 min to stop all virtual machines and calm down it is not helpful when UPS kills power after 2 min.
> 
Understood. We try to collect these differences between spec and response in the DDL, but your UPS had not been reported as of yet: http://buildbot.networkupstools.org/~buildbot/cayman/docs/latest/ddl/ (canonical URL is http://www.networkupstools.org/ddl/ but I'm getting a 503 error at the moment.)

There is a similar problem with this UPS: http://buildbot.networkupstools.org/~buildbot/cayman/docs/latest/ddl/Fideltronik_INIGO/Viper_1200.html (Not to pick on Fideltronik, but they were the first entry I found.)

-- 
Charles Lepple
clepple at gmail