Monitoring Your UPS With apcupsd
In late 1996, Andre M. Hedrick started a Linux project called apcupsd, a dæmon whose purpose was to monitor APC UPSes for power alerts and to shut down the computer when needed. Development of this software was possible because of information gathered by the Internet, and by directly analyzing APC's protocols.
I joined the project in October 1997 after having bought a Smart UPS v/s 650 that I still have. However, due to old age, its battery is now dead. (Many thanks to APC who gave me a new Smart UPS 1400INET, allowing me to continue development of apcupsd.)
From the beginning, apcupsd had been licensed under the GNU Public License (GPL). The GPL license was chosen because apcupsd was intended to be software for anyone, with full sources distributed without charge, and with the best support its developers were able to give.
In mid 1998, legal issues raised by APC forced Andre to remove apcupsd from its public place and distribute it as a binary-only package, removing the GPL license from its source code. This caused a lot of discussion in the Internet community during the following year.
On April 7, 1999, APC withdrew its legal objections and started actively helping our team. This allowed apcupsd to return to its original license. Nowadays, APC monitors our project on our development mailing-list and gives help on technical issues.
In September 1999, Kern Sibbald joined the project. Being an experienced software developer, he quickly became one of the main apcupsd developers.
apcupsd's main task is to monitor the UPS status continuously, and take action based on information received from it. Of course in real life it is not so simple.
apcupsd must be run at startup time, when the operating system services are loaded: in fact, apcupsd is just another OS service. Typically, apcupsd is run as a dæmon (i.e. in the background) with root privileges, in order to be able to take the actions needed to keep the computer healthy. Usually, it is run by the system startup scripts when the system goes multiuser.
Because of its tight relationship with the OS, the source tree contains automated installation of startup and shutdown scripts. During the compilation stage the apcupsd, initialization and control scripts are customized for the local OS. Let's see which system files are modified, and which new files are installed by apcupsd's installation process in a SuSE Linux system.
apcupsd's main init script file is installed in:
This script is responsible for starting up apcupsd during system startup and shutting down apcupsd during system shutdown. It is also symbolically linked to these paths:
/sbin/init.d/rc2.d/K20apcupsd /sbin/init.d/rc2.d/S20apcupsd /sbin/init.d/rc3.d/K20apcupsd /sbin/init.d/rc3.d/S20apcupsdThey are present only on runlevel 2 and 3 because apcupsd is run only in multiuser runlevels; that means runlevel 2 or 3 on SuSE Linux OS.
To be able to shutdown the computer properly on power failures, apcupsd relies on its own service script located in:
and on a patched halt script. When apcupsd detects a situation that needs an emergency shutdown, it first creates two files called /etc/nologin and /etc/apcupsd/powerfail.
Then it initiates the system shutdown. apcupsd will be killed by the system during this phase. The no login file is needed to inhibit user logins during emergency, while the power-fail file is needed during system shutdown. It is used by /sbin/init.d/halt.local as shown in Listing 1. This script will be run during system shutdown after the main halt script is run. When shutting down a computer connected to a failing UPS, we must make sure that UPS power will be removed well after all processes have been killed, and the system disks have been unmounted. In order to do this, apcupsd installation modifies the halt.local file to execute the needed actions. If the power-fail file is present during a system shutdown, all processes are killed, then all the local disks are remounted read-only, and finally we send a power shutdown command to the UPS. The final effect is that, first of all, the system is put in a safe state, and only then is the UPS power is switched off.
The computer is switched off as well by lack of power, but its main switch button is left in the ON position. When utility power returns, the UPS will, after a delay to make sure the power is stable, switch on the output power, and the computer will be restarted automatically. No user intervention is needed during these operations.
|Android Candy: Intercoms||Apr 23, 2015|
|"No Reboot" Kernel Patching - And Why You Should Care||Apr 22, 2015|
|Return of the Mac||Apr 20, 2015|
|DevOps: Better Than the Sum of Its Parts||Apr 20, 2015|
|Play for Me, Jarvis||Apr 16, 2015|
|Drupageddon: SQL Injection, Database Abstraction and Hundreds of Thousands of Web Sites||Apr 15, 2015|
- "No Reboot" Kernel Patching - And Why You Should Care
- DevOps: Better Than the Sum of Its Parts
- Android Candy: Intercoms
- Return of the Mac
- Drupageddon: SQL Injection, Database Abstraction and Hundreds of Thousands of Web Sites
- Designing Foils with XFLR5
- Non-Linux FOSS: .NET?
- Play for Me, Jarvis
- Consent That Goes Both Ways