[lug] Strange Red Hat experience
stimits at idcomm.com
Mon Jul 29 17:56:50 MDT 2002
Peter Hutnick wrote:
> I just lost power to my home "server" and when it came back up I couldn't
> connect via SSH or HTTP. I could ping, so I knew that it was back up.
> Sooo, I drug the monitor up from the basement and plugged it in.
> Everything looked fine. Restarted services, fsck, yada, yada. Still no
> Finally I look at what daemons are running, and damned if ipchains and
> iptables weren't running. I don't run them normally, and don't have them
> configured for my environment (i.e. to let SSH and HTTP through). (Please
> send chiding emails to /dev/null) WTF? I have had other power failures
> and every thing has come up roses in the past.
> Can anyone offer a rational explanation?
> Web Page: http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> Join us on IRC: lug.boulder.co.us port=6667 channel=#colug
What filesystem types are involved? I will assume Redhat for filesystem
structure, probably the parts I am thinking of are similar layout for
other distros, YMMV.
If lock files are left over (check /var/lock/subsys/), then a startup of
a service such as ipchains or httpd could fail (those are just for sake
of argument, I do not know how their scripts deal with stale locks). In
particular, postgresql is this way, at least on RH 7.1. I had to rewrite
parts of the init script to deal with daemons not active but subsys
locked. To see this, one had to cd to /etc/rc.d/init.d/ (use whichever
init.d directory your system uses), and run "./postgresql status". When
it says service is stopped but subsystem is locked, you need to go to
/var/lock/subsys/, and remove the stale lock, then run "./whatever start".
D. Stimits, stimits AT idcomm.com
More information about the LUG