HP3000-L Archives

July 1997, Week 4

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Reply To:
Date:
Wed, 23 Jul 1997 13:24:10 -0700
Content-Type:
text/plain
Parts/Attachments:
text/plain (48 lines)
Item Subject: cc:Mail Text
Informix:

I was building a clustered index on a 3.8M record table in an Informix
database under HP-UX, and the process had been running for about a half
hour, when I absentmindedly launched a program that prepared both SELECT
and DELETE statements against this table. The statements would never have
been executed, but they were PREPAREd.

You know how when you press return and immediately everything stops, you
*know* in your gut that you just caused a system failure? Well, it wasn't
that extreme, but not only did the app fail with an error (as expected
under the circumstances), but the clustering process also stopped with an
error. I tried to open the database, but could not. The "database instance"
(which controls several databases) had to be brought down and back up by
the database system administrators when they came in this morning.

Based on this and other events, I conclude that Informix is powerful, but
both fragile and resource-intensive. Any other experiences/tips?


Image:

Working on an HP-UX workstation under X-windows, a project manager had
windows open for sessions on both the development 3000 and the production
3000. She wanted to restore a database into the development account. You
guessed it -- she was in the wrong window, and started restoring over the
production database. (I don't know the logistics of how this happened, only
that networking is extensive, and it happened "somehow.") Some users were
still on, so some sets could not be restored, while others could.

Fortunately, logging is used on that database. She restored from the
morning backup, then applied roll-forward recovery until about the time
that she began the restore. The database is hit pretty heavily during the
day, and the logfiles totaled around 1.7M records, but it finished in
about an hour and a half. No problem, though the documentation on
roll-forward recovery could be more clear.

Based on this and other events, I conclude that Image is powerful, robust,
and resource-friendly.


FWIW.

--Glenn Cole
  Software al dente, Inc.
  [log in to unmask]

ATOM RSS1 RSS2