NUT package (2.8.0 and below)

PatRyan

Update - No issues overnight still running fine and serving my network as a NUT Server. Updated to the latest 2.7.0 Devel build this morning and UPS service started up without a hiccup. Very pleased.

FYI, I have a CyberPower CP1500PFCLCD and it is connected to my pfSense box via the supplied USB cable.

OffstageRoller

@dennypage said in NUT package:

I've received several requests for the dev build of usbhid-ups, so I thought I would upload the file here.

usbhid-ups.gz

For reference, the shasum and sha256sum checksums of the unzipped file are:
49ce9131502bfb8b789ee97b7fb3fc81fc9f8fff  usbhid-ups
999a2653559dbc50ecc8ba592a67587b1e307a1495f6e8ebbd3d8e90e3967133  usbhid-ups
If you use the file, please post and let me know if it resolves an issue for you.

So far so good for me with a Trip Lite SMART1500LCD! It's only been an hour, but it has stayed connected and my logs are no longer getting spammed by disconnects/connects.

OffstageRoller

@offstageroller said in NUT package:

@dennypage said in NUT package:
I've received several requests for the dev build of usbhid-ups, so I thought I would upload the file here.

usbhid-ups.gz

For reference, the shasum and sha256sum checksums of the unzipped file are:
49ce9131502bfb8b789ee97b7fb3fc81fc9f8fff  usbhid-ups
999a2653559dbc50ecc8ba592a67587b1e307a1495f6e8ebbd3d8e90e3967133  usbhid-ups
If you use the file, please post and let me know if it resolves an issue for you.
So far so good for me with a Trip Lite SMART1500LCD! It's only been an hour, but it has stayed connected and my logs are no longer getting spammed by disconnects/connects.

I'm at about 8 hours now, with no issues to report. All is well again with my USB connection to my UPS!

knight-of-ni

Hey guys. Mind if I join the party?

I upgraded from from 2.6 CE to 23.01 plus today, to get support for the 2.5Gbps nics in my firewall.

Unfortunately after the upgrade, NUT started failing because it couldn't claim the USB device:

Can't claim USB device [051d:0003]@0/0: Other error

UPS is a APC Smart-UPS 1500.

Did a little searching, found it was a permission error, and eventually found this thread.

Looks like I found the right place.

I've gone back through this thread about a month and started reading.

Adding user=root to ups.conf got things going again. However, I'd call that a workaround. If I read right, looks like I need to wait for the next release of NUT for a real fix.

At the moment I am not using @dennypage's custom usbhid-ups. If my UPS does not stay online, I'll apply it and post results.

I'll be keeping an eye on this thread for new information.

dennypage

@knight-of-ni said in NUT package:

Adding user=root to ups.conf got things going again. However, I'd call that a workaround. If I read right, looks like I need to wait for the next release of NUT for a real fix.

While the next release of nut is expected to address the CyberPower issue, it will not address the APC issue. The APC issue is a usb quirk issue, and this requires a new version of the kernel in pfSense to permanently resolve. I don't expect that to happen soon.

See here for further details of the APC issue. I recommend using the /boot/loader.conf.local solution if you can take a reboot.

dem

@dennypage said in NUT package:

If you use the file, please post and let me know if it resolves an issue for you.

Your version of usbhid-ups has been working fine with my CyberPower for 48 hours now. Thanks.

jwfox5150

@dennypage said in NUT package:

If you use the file, please post and let me know if it resolves an issue for you.

I was experiencing the same issue and using this version of usbhid-ups has resolved the issue. Many thanks, that was driving me NUTS (pun intended)

JeGr

@dennypage Quick question: is the NUT package somehow dependent on other services or is integrated into a service hook since 23.01? After upgrading I have NUT alert messages all over the place for things that are completely unrealted to the UPS.
E.g. resetting or reconfiguring a gateway, doing configurations on WAN or VPNs if the interface is assigned etc. all seem to trigger "problems" with NUT loosing and regaining connection to the UPS. UPS attached is an APC BackUPS via USB that ran well before without being trigger happy with notifications when interface/gateway/routing things are happening. Now just touching some of those things seem to trigger a connection loss from NUT. Really confusing

Cheers
\jens

dennypage

@jegr said in NUT package:

E.g. resetting or reconfiguring a gateway, doing configurations on WAN or VPNs if the interface is assigned etc. all seem to trigger "problems" with NUT loosing and regaining connection to the UPS.

pfSense restarts package services when WAN interfaces disconnect or reconnect. Yes, this is unnecessary for some services, such as NUT with a USB connection, but there is no way for pfSense to know which services actually need to be restarted. It's always been this way.

What you would expect to see is NUT restart once when the interface goes down, and once again when the interface comes back up. Whether or not you see NUT reporting a lost connection or not depends upon the order and speed of shutting down the various processes involved (usbhid-ups, upsd, upsmon).

JeGr

@dennypage said in NUT package:

pfSense restarts package services when WAN interfaces disconnect or reconnect. Yes, this is unnecessary for some services, such as NUT with a USB connection, but there is no way for pfSense to know which services actually need to be restarted. It's always been this way.

That may very well be, but before 23.01 there were no problems with NUT overactively reporting down/ups at those times whereas now they pop up almost every time when someone is editing something on interfaces, routing gateways etc.
Just wanting to check if there's anything that has changed while converting stuff to PHP 8.1 or anything. Wouldn't be the first :)

dennypage

@jegr Nothing to do with PHP, it's actually a shell script.

If you go back through this thread you'll find mention of this issue five or more years ago. Some people see it a lot, some never see it. It just depends on how things happen on the box. If it is really bothering you, you could do a local patch for the rc script to add a sleep following the kill of upsmon. But only if it's really bothering you.

JeGr

@dennypage Nah, I can live with that :) But that only adds to another thing that packages should have more specific hooks like the XY interface instead of "any" interface, as theres no need otherwise. Or OpenVPN that gets restarted pretty often while listening on localhost so would be completely unphased by any changes. Also it'd be nice to have those notifications selectable and expandable so we can get more/specific notifications and disable others :)

But I'll push that in another thread. Thanks for getting back :)

Cheers
\jens

DavidIr

@dennypage said in NUT package:

I've received several requests for the dev build of usbhid-ups, so I thought I would upload the file here.

usbhid-ups.gz

For reference, the shasum and sha256sum checksums of the unzipped file are:
49ce9131502bfb8b789ee97b7fb3fc81fc9f8fff  usbhid-ups
999a2653559dbc50ecc8ba592a67587b1e307a1495f6e8ebbd3d8e90e3967133  usbhid-ups
If you use the file, please post and let me know if it resolves an issue for you.

I finally got round to loading this file onto my system, and unfortunately it does not seem to work at all. Now instead of failing randomly after a period of time I get the failure immediately after restarting the UPS service.


Mar 30 18:25:59	upsmon	34059	Poll UPS [EatonUPS] failed - Driver not connected
Mar 30 18:25:54	upsmon	34059	Poll UPS [EatonUPS] failed - Driver not connected
Mar 30 18:25:49	upsmon	34059	Poll UPS [EatonUPS] failed - Driver not connected
Mar 30 18:25:44	upsmon	34059	Poll UPS [EatonUPS] failed - Driver not connected
Mar 30 18:25:39	upsmon	34059	Poll UPS [EatonUPS] failed - Driver not connected
Mar 30 18:25:38	php-cgi	35430	notify_monitor.php: Message sent to my@email.net OK
Mar 30 18:25:34	upsmon	34059	Poll UPS [EatonUPS] failed - Driver not connected
Mar 30 18:25:29	upsmon	34059	UPS EatonUPS is unavailable
Mar 30 18:25:29	upsmon	34059	Poll UPS [EatonUPS] failed - Driver not connected
Mar 30 18:25:29	upsd	35561	User local-monitor@::1 logged into UPS [EatonUPS]
Mar 30 18:25:27	php-cgi	35430	notify_monitor.php: Message sent to my@email.net OK
Mar 30 18:25:25	upsd	35561	Startup successful

I think this log sequence shows from startup.

Any further suggestions for a solution to this? Are there any diagnostics I can do to be helpful?
Thank you.
David

PatRyan

@davidir did you check to be sure the file permissions are correct after uploading it? This file from Denny Page is working without a hitch for me with a CyberPower UPS. I don't know if there are any other quirks with an Eaton UPS.

dennypage

@davidir said in NUT package:

I think this log sequence shows from startup.
Any further suggestions for a solution to this?

What's missing from your log entries are the usbhid-ups entries. Those are what is needed to help further the investigation if you are having a problem with usbhid-ups.

You may want to confirm that it is installed correctly. Location and permissions should look like this:

[23.01-RELEASE][root@fw]/root: ls -l /usr/local/libexec/nut/usbhid-ups 
-rwxr-xr-x  1 root  wheel  250968 Jan 17 17:34 /usr/local/libexec/nut/usbhid-ups
[23.01-RELEASE][root@fw]/root:

Lastly, if you are still having an issue you may want to post your config.

DavidIr

@dennypage Just to be clear (in case I did something really stupid) here is what I have done:

download file linked above 1678659799995-usbhid-ups.gz
Use pfsense web /diag_command.php to upload file

Uploaded file to /tmp/1678659799995-usbhid-ups.gz.

Stop UPS service on pfsense
use putty to open shell on pfsense
check existing file

[23.01-RELEASE][admin@pfSense.irwazu.co.uk]/usr/local/libexec/nut:  ls -l /usr/local/libexec/nut/usbhid-ups
-rwxr-xr-x  1 root  wheel  202096 Jan 18 01:29 /usr/local/libexec/nut/usbhid-ups
[23.01-RELEASE][admin@pfSense.irwazu.co.uk]/usr/local/libexec/nut:
[23.01-RELEASE][admin@pfSense.irwazu.co.uk]/usr/local/libexec/nut:

extract the .gz file downloaded

[23.01-RELEASE][admin@pfSense.irwazu.co.uk]/usr/local/libexec/nut: gunzip -d /tmp/1678659799995-usbhid-ups.gz
[23.01-RELEASE][admin@pfSense.irwazu.co.uk]/usr/local/libexec/nut: ls -l /tmp/1678659799995-usbhid-ups
-rw-r--r--  1 root  wheel  258936 Mar 31 17:08 /tmp/1678659799995-usbhid-ups
[23.01-RELEASE][admin@pfSense.irwazu.co.uk]/usr/local/libexec/nut:

rename old file, move new file into place

[23.01-RELEASE][admin@pfSense.irwazu.co.uk]/usr/local/libexec/nut: mv usbhid-ups usbhid-ups.old
[23.01-RELEASE][admin@pfSense.irwazu.co.uk]/usr/local/libexec/nut: mv /tmp/1678659799995-usbhid-ups ./usbhid-ups

fix new file permissions

[23.01-RELEASE][admin@pfSense.irwazu.co.uk]/usr/local/libexec/nut: chmod 0755 usbhid-ups
[23.01-RELEASE][admin@pfSense.irwazu.co.uk]/usr/local/libexec/nut: ls -l /usr/local/libexec/nut/usbhid-ups
-rwxr-xr-x  1 root  wheel  258936 Mar 31 17:08 /usr/local/libexec/nut/usbhid-ups

start the UPS service

Config:
no additional driver arguments
Additional configuration lines for upsmon.conf: RUN_AS_USER root
Additional configuration lines for ups.conf: user=root
Additional configuration lines for upsd.conf:
LISTEN 127.0.0.1
LISTEN 192.168.200.254

These are there as I want to be able to access the UPS status from an external monitor, and this was one of the steps in the documentation I followed.

I am still not seeing any log lines from usbhid-ups, if I revert back to the original usbhid-ups file then these do show.

dennypage

@davidir said in NUT package:

Additional configuration lines for upsmon.conf: RUN_AS_USER root
Additional configuration lines for ups.conf: user=root

FWIW, you should remove both of these lines. You don't want to run as root unless you really have to. If you are running as root to work around a missing quirk issue, I strongly recommend using the loader.conf.local approach instead.

Setting that aside...

You can try logging in to your box and running the driver by hand to confirm, but I suspect that I already know why it isn't working for you. The executable I posted is for 64-bit x86-64 (Intel or AMD) hardware... what hardware architecture are you on?

The reason I ask, is that my files look like so:

[23.01-RELEASE][root@fw]/root: ls -l /usr/local/libexec/nut/usbhid-ups*
-rwxr-xr-x  1 root  wheel  258936 Apr  1 00:26 /usr/local/libexec/nut/usbhid-ups
-rwxr-xr-x  1 root  wheel  250968 Jan 17 17:34 /usr/local/libexec/nut/usbhid-ups.org
[23.01-RELEASE][root@fw]/root:

My original usbhid-ups is 250,968 bytes, whereas yours is only 202,096 bytes. This suggests a different architecture. Are you on an ARM by chance?

DavidIr

Yes I have a Netgate 3100, CPU Type: ARM Cortex-A9 r4p1

@dennypage said in NUT package:

FWIW, you should remove both of these lines. You don't want to run as root unless you really have to. If you are running as root to work around a missing quirk issue, I strongly recommend using the loader.conf.local approach instead.

Thanks for the pointer, I will look into this.

stompro

Thank you @dennypage , I just used the described method to fix my Tripp Lite SMART750RM1U using the tripplite_usb driver with pfSense 23.01 running on a 6100.

usbconfig -d ugen0.2 dump_device_desc
ugen0.2: <Tripp Lite TRIPP LITE SMART750RM1U> at usbus0, cfg=0 md=HOST spd=LOW (1.5Mbps) pwr=ON (0mA)

  bLength = 0x0012
  bDescriptorType = 0x0001
  bcdUSB = 0x0110
  bDeviceClass = 0x0000  <Probed by interface class>
  bDeviceSubClass = 0x0000
  bDeviceProtocol = 0x0000
  bMaxPacketSize0 = 0x0008
  idVendor = 0x09ae
  idProduct = 0x0001
  bcdDevice = 0x000a
  iManufacturer = 0x0001  <Tripp Lite >
  iProduct = 0x0002  <TRIPP LITE SMART750RM1U >
  iSerialNumber = 0x0000  <no string>
  bNumConfigurations = 0x0001

#Two lines shown
usbconfig -d ugen0.2 show_ifdrv
ugen0.2: <Tripp Lite TRIPP LITE SMART750RM1U> at usbus0, cfg=0 md=HOST spd=LOW (1.5Mbps) pwr=ON (0mA)
ugen0.2.0: uhid0: <Tripp Lite TRIPP LITE SMART750RM1U, class 0/0, rev 1.10/0.0a, addr 2>

#boot time config
hw.usb.quirk.0="0x09ae 0x0001 0x0000 0xffff UQ_HID_IGNORE"
/boot/loader.conf.local

#live testing
usbconfig add_dev_quirk_vplh 0x09ae 0x0001 0x0000 0xffff UQ_HID_IGNORE

stompro

Is the nut pfsense package setup to deal with power race conditions?

I'm trying to understand when /usr/local/etc/rc.d/shutdown.nut.sh gets executed.

My UPS will cancel a delayed shutdown if the power returns during the shutdown delay, so that would leave my firewall in a halted state, never to restart.

So should a shutdown delay ever be used with the pfsense nut package? Or is it best to shutdown the UPS immediately when "/usr/local/sbin/upsdrvctl shutdown" gets called?

The nut FAQ has a recommendation for adding in a reboot after 120 seconds if the system is still up after the upsdrvctl shutdown command has been run.

https://networkupstools.org/docs/FAQ.html#_i_8217_m_facing_a_power_race