jimklimov
commented
2 weeks ago Hello, thanks for the interesting question.
First of all, please note that you seem to be running NUT v2.7.2 which is over 10 years old and would not be directly updated, at least by core project. You might build your own if required, though (e.g. to apply a custom patch). Current NUT codebase aims to be buildable wherever it worked before, so https://github.com/networkupstools/nut/wiki/Building-NUT-for-in%E2%80%90place-upgrades-or-non%E2%80%90disruptive-tests might help you follow that path.
Looking at even current code, the command sequence in https://github.com/networkupstools/nut/blob/d244d73b1c90d5a461b39023bd4900f9eacd935d/drivers/usbhid-ups.c#L969-L993 is hard-coded. I suppose a new driver configuration variable to list the suitable shutdown commands in order can make sense here.
Note that normally NUT daemons (drivers, upsmon, upsd...) are killed as any other services in the system going down, before it gets to the part where it calls upsdrvctl shutdown -> driver -k, so you would not be able to simply upscmd the needed operation. Generally it depends on your OS and its NUT integration, in some cases you might be able to modify the shutdown sequence so that NUT daemons survive until power-cycling. In others, the OS kills off all remaining processes before the endgame.
Dummy0815
Hi we are using an APC USB UPS with an Try Contact I/O card (AP9613) with an external key switch, to switch the UPS and therefor the connected systems with this key ON/OFF. The APC is "simulating" an LOW BAT with key position OFF and triggers a UPS ON with the other position. Everything works fine with the 'old' generation of APC (SUA750RMI1U, SUA750I, ..) but with the newer generation (SMT1500RMI2U) we have the issue, that if key OFF (low bat) triggers the system shut down, but the last command "upsdrvctl shutdown -> driver -k" sends an "shutdown.reboot" to the UPS. This causes the UPS to turn the outlets delayed off (which is what we want), but issuing a restart, which causes the UPS to turn on the outlets again (as main power is on the the whole time). Reason seems to be that the 'old' APC's support the shutdown commands: "shutdown.reboot", "shutdown.return", "shutdown.stayoff" (and load.off.delay) While the newer ones only support: "shutdown.reboot" (and load.off.delay)
So according the NUT documentation the driver -k tries to issue in a fixed order "shutdown.return", "shutdown.reboot", "load.off.delay". So old APC is shutdown with "shutdown.return" which is Ok as the UPS stays down as the main power doesn't change. While the newer, as they down support the first command, are doing an "shutdown.reboot" causing the issue we have.
So my question is there any parameter which influences the attempted command sequence? So that we could configure try: 'load.off.delay' first than, "shutdown.return", than "shutdown.reboot"? As an workaround is there any chance to send a 'load.off.delay' in the final system shutdown stage instead of "upsdrvctl shutdown"?
For reference the output of an old vs an newer APC: ######################## upsc shups .. device.mfr: American Power Conversion device.model: Smart-UPS 750 driver.name: usbhid-ups driver.version.data: APC HID 0.95 driver.version.internal: 0.38
upscmd -l shups .. load.off.delay - Turn off the load with a delay (seconds) shutdown.reboot - Shut down the load briefly while rebooting the UPS shutdown.return - Turn off the load and return when power is back shutdown.stayoff - Turn off the load and remain off
######################## upsc shups .. device.mfr: American Power Conversion device.model: Smart-UPS 1500 driver.name: usbhid-ups driver.version: 2.7.2 driver.version.data: APC HID 0.95
upscmd -l shups .. load.off.delay - Turn off the load with a delay (seconds) shutdown.reboot - Shut down the load briefly while rebooting the UPS