[CentOS] Power Fail Protection

Wed Aug 9 00:07:43 UTC 2017
Valeri Galtsev <galtsev at kicp.uchicago.edu>

On Tue, August 8, 2017 6:50 pm, Chris Olson wrote:
> Some of our largest systems run Windows because it supports engineering
> applications that we use regularly.  These applications have unattended
> runs that often take between ten and fifteen hours to complete.  We have
> taken the recommendation of the application supplier and equipped these
> Windows machines with UPS protection for 30 minutes at full load.
> The UPSs are Ethernet connected.  A support application on the Windows
> engineering machine communicates with the UPS to detect and address any
> facility power failure.  The long run engineering application is then
> suspended at a restart point and the system is shut down.  We initiate
> job completion manually from the suspension restart point after the
> system has reliable power and is rebooted.
> If we wanted to protect our CentOS systems from facility power failure
> in a similar way, is there operating system or other standard support
> that we might employ?  Most of the Linux-based applications are not as
> critical as the engineering applications on the Windows machines. There
> is a significant amount of processor idle time on several of the CentOS
> systems during non-work hours when the systems are unattended.  Several
> CentOS systems are supported currently with UPSs, but they run out and
> the system loses power if it is unattended.

I used a lot APC smart UPSes. They have serial or USB connection through
which some daemon you run on your machine (apcupsd) can detect UPS on
battery, and can initiate clean system shutdown when battery is below some
charge level (you define which in configuration). Apcupsd is free open
source software, I never used APC's software that does the same. When
power returns (if UPS fully drained its battery) system can be configured
to boot on power restored. If you have more than one machine behind the
same UPS, apcupsd daemons on other machines can run in "slave" mode and
get information from master apcupsd.

Depending on your UPS make/model there may be similar daemon that can do
the same.


> _______________________________________________
> CentOS mailing list
> CentOS at centos.org
> https://lists.centos.org/mailman/listinfo/centos

Valeri Galtsev
Sr System Administrator
Department of Astronomy and Astrophysics
Kavli Institute for Cosmological Physics
University of Chicago
Phone: 773-702-4247