> How do UNIX/Solaris systems respond if a UPS supply
> runs out of battery power and no one is available to
> manually shut them down?
> I know there is software to automate this, but it is
> not currently available at this site and the UPS's are.
> I've got several unix boxes including 2 solaris systems
> running full time. Our electrical supply is bad
> enough that we probably have 15 - 20 outages a year.
> Some in short order during storms; on/off/on/off ...
> Fortunately, I have never noted a data loss or file
> system corruption due to these failures. But it is
> just a matter of time. It is also funny how the
> various systems are different in their sensitivity
> to power "flickers".
We've found Solaris much better in this respect than many
other unixs we use. We also suffer typically 20 outages
a year (generally the UK is does not have this problem,
but our little village does). I am refering to Solaris
x86, so sensitivity to power "flickers" depends on PC's
I find that, unless a disc was actually being written to
the instant the power went, it seems to come back up as
'stable' (rather than 'clean') and seems not to require
fscking. Discs that were being written to do seem to be
fscked. However, I've never yet had a corruption due to
power failure which fsck could not fix, and I've had lots
of power failures! I do keep very good backups though.
(System is an NFS and Samba server for some 40 clients.)
We now have the system on a UPS, but there is no UPS
protection for any of the means to access the system.
Thus, the system has normally been quiet for some time
when the UPS battery dies. The result is that all the
filesystems come back up as 'stable' and don't even
I also don't configure the UPS software to shut down
the system just before battery runs out, because there
is a bug in it. (If the power comes back whilst the
system is shutting down, it still continues to shut
down and stays down; this is unacceptable to us.)
We don't run ODS or prestoserve.