Published: 2013-02-27

This post explains how to configure Network UPS Tools (NUT) to work with an Eaton 3S UPS on a stand-alone Ubuntu 12.04 PC.

Network UPS Tools (NUT) seems to be the most widely supported and used UPS management package for Linux. It is extremely flexible and caters for a wide range of UPS makes, models and deployment scenarios. The sheer scope and flexibility of NUT makes UPS selection and configuration daunting. The purpose of this article is to illustrate a minimal configuration to achieve stand-alone UPS management of a single Ubuntu 12.04 based PC.

NUT configuration

Here’s how this configuration works: when power has dropped out for more than 3 minutes NUT triggers a shutdown. The shutdown sequence closes down Ubuntu; commands the UPS to commence shutdown; and then turns off the PC. Once the PC is off the UPS turns off to conserve the UPS battery. When power is reestablished the UPS powers up and reapplies power to the PC (whether your PC turns on automatically at power resumption is determined by it’s BIOS settings).

Here are the minimal NUT configuration files:

/etc/nut/nut.conf

# Set MODE=none to disable UPS monitoring, MODE=standalone to enable UPS monitoring.
MODE=standalone

/etc/nut/ups.conf

[eaton3s]
driver=usbhid-ups
port=auto

/etc/nut/upsd.conf

/etc/nut/upsmon.conf

MONITOR eaton3s@localhost 1 monuser pass master
SHUTDOWNCMD "/sbin/shutdown -P now"
POWERDOWNFLAG /etc/killpower

NOTIFYFLAG ONBATT SYSLOG+WALL+EXEC
NOTIFYFLAG ONLINE SYSLOG+WALL+EXEC
NOTIFYCMD "/etc/nut/notifycmd"

/etc/nut/upssched.conf

/etc/nut/upsd.users

[monuser]
password=pass
upsmon master

The following notifycmd bash script handles NUT ONBATT and ONLINE events. Put it in /etc/nut, it is executed when NUT detects power resumption (prior to the 3 minute timeout) and when NUT detects the UPS has switched to battery (power outage). Don’t forget to make this script executable (sudo chmod +x /etc/nut/notifycmd).

/etc/nut/notifycmd

#!/bin/bash
#
# NUT NOTIFYCMD script

PATH=/sbin:/usr/sbin:/bin:/usr/bin:/usr/local/sbin:/usr/local/bin

trap "exit 0" SIGTERM

if [ "$NOTIFYTYPE" = "ONLINE" ]
then
        echo $0: power restored | wall
        # Cause all instances of this script to exit.
        killall -s SIGTERM `basename $0`
fi

if [ "$NOTIFYTYPE" = "ONBATT" ]
then
        echo $0: 3 minutes till system powers down... | wall
        # Loop with one second interval to allow SIGTERM reception.
        let "n = 180"
        while [ $n -ne 0 ]
        do
                sleep 1
                let "n--"
        done
        echo $0: commencing shutdown | wall
        upsmon -c fsd
fi

Testing the UPS

• Check NUT status:
  

  $ sudo service nut status Checking status of Network UPS Tools * upsd is running * upsmon is running

• If necessary start NUT:
  

  $ sudo service nut start

• Plug in USB cable and check it’s been detected:
  

  $ lsusb : Bus 005 Device 003: ID 0463:ffff MGE UPS Systems UPS

• Check the UPS status:
  

  $ sudo upsc eaton3s battery.charge: 100 battery.charge.low: 20 battery.runtime: 3000 battery.type: PbAc device.mfr: EATON device.model: Eaton 3S 700 : ups.status: OL CHRG ups.timer.shutdown: -1 ups.timer.start: -1 ups.vendorid: 0463

• Check power off/on status by unplugging the power to the UPS until it beeps a few times then plug it back in. You will get the following console messages:
  

  Broadcast Message from nut@nas1 (somewhere) at 13:23 ... UPS eaton3s@localhost on battery Broadcast Message from nut@nas1 (somewhere) at 13:23 ... /etc/nut/notifycmd: 3 minutes till system powers down... Broadcast Message from nut@nas1 (somewhere) at 13:23 ... UPS eaton3s@localhost on line power Broadcast Message from nut@nas1 (somewhere) at 13:23 ... /etc/nut/notifycmd: power restored

• The /var/log/syslog should have two messages like:
  

  Feb 19 11:37:54 nas1 upsmon[3044]: UPS eaton3s@localhost on battery Feb 19 11:38:14 nas1 upsmon[3044]: UPS eaton3s@localhost on line power

• Simulate a power outage (NOTE: this will take the PC to shutdown immediately):
  

  $ sudo upsmon -c fsd

• Test the /etc/nut/notifycmd script fully by unplugging the power from the UPS and waiting until the full shutdown sequence is played out (takes 3 minutes). Look for the message:
  

  /etc/nut/notifycmd: 3 minutes till system powers down...

UPS selection

The key to using a UPS with NUT is getting a UPS that is fully supported by the NUT version you are using (Ubuntu 12.04 has NUT 2.6.3). Here’s why I chose the Eaton 3S:

1. It is fully supported by NUT UPS Tools 2.6.3 on Unbuntu 12.04 with the usbhid-ups driver.
2. The vendor has a Linux commitment — they provide their own Intelligent Power Protector (IPP) software for Linux (I didn’t use it, choosing NUT instead).
3. Readily available replacement battery.
4. Other users report it works with Ubuntu 12.04.

I was faced with the problem of cloning one Ubuntu server box to another. My solution is purposefully minimalist, requiring only a bootable Linux CD-ROM (I used the Xubuntu 10.04 Desktop CD-ROM) and an external USB drive to transfer the HDD data.

The procedure involves backing up and restoring the MBR, partition table and partition image files then manually tweaking hardware configuration differences. The examples are for HDD /dev/sda with a single ext3 partition and external USB backup drive /dev/sdb with an ext3 partition /dev/sdb1 — you need to adjust the drive and partition device names to match your environment.

• The MBR is reinstalled by the GRUB boot loader instead of backing up and restoring the raw 512 byte MRB sector because the existing MBR will only work if the target HDD has exactly the same geometry as the source drive.
• The target HDD’s geometry doesn’t need to match the source drive but the target drive’s capacity cannot not be less that the source drive. You can use hdparm(1) to view the drive’s parameters:

  hdparm -gi /dev/sda

• Because raw partitions are involved this procedure should also work for non-Linux operating systems, but you will need to use the tools recommended by your operating system to regenerate the MBR and tweak the hardware configuration differences.
• This procedure can be used to upgrade to a larger HDD. After you’ve restored to the new disk run gparted and resize your primary partition (on the Xubuntu 10.04 Live CD you will find gparted in the Applications→System menu).
• Most of the commands require superuser privilege — logon as root or apply a sudo command prefix.

Backup

1. Mount the USB backup drive:

   mount /dev/sdb1 /mnt

1. Backup partition table:

   sfdisk -d /dev/sda > /mnt/sda-partition-table

2. Backup partition — the partition must be unmounted to copy it so at this point you need to running off the Live CD.
3. Copy the partition to a file sda1-image on the USB drive:

   dd if=/dev/sda1 of=/mnt/sda1-image bs=1K

   Instead of the last command you could reduce the partition size by first zero filling unused space and then compressing with gzip(1):

   dd if=/dev/zero of=zerofill bs=1M
   rm zerofill
   dd if=/dev/sda1 bs=1K | gzip -c > /mnt/sda1-image.gz

Restore

1. Go to the target PC and boot the Live CD.
2. Mount the USB backup drive:

   mount -o ro /dev/sdb1 /mnt

3. Restore partition table:

   sfdisk /dev/sda < /mnt/sda-partition-table

   Note	If there was already a swap partition on the target HDD the Live CD may be using it and you will get a device busy message (you can check by running the swapon -s command). Use swapoff(8) to umount the swap, for example: swapoff /dev/sda5

1. Restore partition:

   dd if=/mnt/sda1-image of=/dev/sda1 bs=1K

   Or if you compressed the partition:

   gzip -cd /mnt/sda1-image.gz | dd of=/dev/sda1 bs=1K

2. Write a new MBR using GRUB. You need to use the version of GRUB (GRUB Legacy or GRUB 2) used by the cloned machine. The easiest way to do this is to change root to the cloned root partition and then run GRUB in it’s native environment:

   mount /dev/sda1 /mnt
   chroot /mnt

   If you have GRUB Legacy (version 0.9x) run these commands:

   grub
   > root (hd0,0)
   > setup (hd0)
   > quit

   If you have GRUB 2 then do:

   grub-install /dev/sda

   You can find some nice Ubuntu GRUB2 documentation here.

3. It’s a good idea to check your restored drive with fsck:

   e2fsck -f /dev/sda1

Updating the target drive

Creating partition images involves taking the source PC down and is time consuming. Instead of maintaining up to date partition images I update the restored target with the latest rsnapshot backups from an external USB drive and use rsync to restore the latest snapshot, for example:

mount -o ro /dev/sdb1 /mnt
mkdir /sda1
mount /dev/sda1 /sda1
rsync -aHxv --delete --exclude /var/local/backups/ \
      /mnt/backups/hosts/kestrel/hourly.0/ /sda1/

The updating is done from the Live CD as it may be undesirable to boot the target machine until it has been completely updated.

Initializing swap space

Before rebooting you need to initialize the target swap partition and update the swap UUID in /etc/fstab. If you don’t initialize the swap it won’t be mounted and you will get boot errors like unable to canonicalize swap (check for mounted swap with swapon -s). Use mkswap(8) to initialize the swap partition, for example:

mkswap /dev/sda5

mkswap(8) will print a UUID, note it down and then edit the swap entry in the target machine’s /etc/fstab file. Alternatively you could just replace the UUID name with the swap device partition name (/dev/sda5 in our example). If your fstab doesn’t use UUIDs you won’t need to change anything. You can print the partition UUID using the vol_id(8) command, for example:

vol_id /dev/sda5

You can now reboot the target machine. If your cloned drive is installed on different hardware you will need to read the next section.

Installing the cloned drive different hardware

Here are post restoration configuration tweaks I had to make in order to get my cloned Ubuntu 7.10 server fully operational.

Even if the target machine is be exactly the same make and model as the original the MAC address will be different. I my case the CD-ROM drive was also different. When I booted the target machine the eth0 network interface failed to come up (check this with ifconfig -a). To regenerate the correct network configuration delete the file

/etc/udev/rules.d/70-persistent-net.rules

and reboot.

Similarly, to regenerate the CD-ROM drive parameters delete the file

/etc/udev/rules.d/70-persistent-cd.rules

then reboot.

The only other configuration problem I had was an unable to iterate IDE devices: missing file or directory message when halting the PC. The problem is benign and is caused by a missing /proc/ide directory (no idea why it’s missing though). The error can be suppressed by removing the halt(8) command -h option in the /etc/init.d/halt script. My guess is that this oddity is distribution dependent.

Finally, don’t forget to exclude the modified configuration files from any subsequent rsnapshot updates you make to the cloned machine.