Re: Quick Tale: Lost production database because of keyboard and Veritas Cluster Server

"RSH" <RSH_Oracle_at_worldnet.att.net> writes:
>
>I forgot about those!! Age is creeping up on me. Well at least I did say
>'something approved by Sun' and 'make sure you test it'... sigh.
>
>Well regardless, (what IS this STOP-A stuff? I never heard it before, what
>is it in ASCII... rut roh, please don't tell me they changed ASCII?!!), the
>poor guy didn't do this on purpose, I don't believe; and you do have to
>admit regardless, that any single signal shouldn't trigger a mess like this
>without some 'are you really sure you want to create weeks of Hell for
>everyone in IT' screen, instead of dropping the system instantly.
>A physical button on the machine itself, okay. But not a signal that can be
>generated so easily and unintentionally, and that can't be redefined by the
>customer and at least be more than one keystroke separating life and death.
>

Never mind that it's lore that's discussed often in these newsgroups and in other places where Sparc administrators congregate, and therefore can be found via Google or other web search sites.

Your friend went up to a server. A server that was not powered down, but up and running. A **production** server.

What did your friend do with this production server? Unplugged hardware from it. The keyboard. Part of ths system console.

The issue here is not why a Sparc might halt when the keyboard is unplugged. That's nowhere NEAR the issue. Neither is the question of KVM switches on Sparc servers.

The issue is, WHAT THE HECK was your friend doing unplugging hardware from a *production* server that was *running*?!? Hardware that's not designated as hot-swappable!

In my opinion, your site needs to implement change control procedures as quickly as you can.

  -Greg

-- 
+++++ Greg Andrews +++ gerg_at_panix.com +++++
I have a map of the United States that's actual size
		 -- Steven Wright