HP3000-L Archives

July 2004, 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:
"Emerson, Tom" <[log in to unmask]>
Reply To:
Emerson, Tom
Date:
Mon, 26 Jul 2004 12:52:09 -0700
Content-Type:
text/plain
Parts/Attachments:
text/plain (35 lines)
> From: Brian Donaldson
> 
> Full system backup to DDS-3 tape takes about an hour [...]
> Partial, nightly backups take about 10 minutes [...]
> 
> [...] store-to-disk runs forever.
> BIG problem. Took over TWO HOURS to do the partial [...]
> over FIVE HOURS to do the full system backup last Friday night.
> 
> I see some things going on that are probably what is doing it in.
> 
> Each disc store file is getting backed up in the current [...]
> So I changed the JCL to whack the previous night's Store disc 
> backup file after the current backup has completed.

The key here is that you're freeing this space AFTER backing it up, thus you are in effect carrying it over day by day -- no suprise that ...
> 
> So this disc file is growing exponentially every day. BIG problem.

Actually, the growth should be linear, or nearly so -- it should be growing by the size of the changed files each day -- so if (typically) 80 meg of data is changed per day, the first day's STD file should be 80 meg; day two it will be 160 meg [80 meg of changes + and an 80 meg backup]; but on the third day it should only be 240 meg, not 320 [80 meg of changes + a 160 meg backup from the previous day]  If you WEREN'T purging the previous day's backup file, then yes, the growth would be a bit more compounded -- day 3 would be 320: 80 meg of changes, an 80 meg backup AND a 160 meg backup...

What you should be doing is placing the store-to-disk files in a particular group or account, then EXCLUDING that group or account from the "backup everything that has changed" script -- then your growth should be marginal [and may even shrink on occaision]

[...]
> What happens if the system goes down and cannot be rebooted 
> from LDEV1 ??
> 
> Oh ok, reload from my trusty CSLT tape. Oh no! Tape drive 
> doesn't work!

This is why people work on things like RAID -- you essentially need a copy of LDEV1 that you can pop-in at a moment's notice.  You probably *could* get away with a base-level install, then pulling that disk and storing it on a shelf while you install on a second drive; this drive then gets used day-by-day; should the need to "recover" arrise, you pull the pre-built copy from the shelf and boot that instead.  You'll lose anything else that was on "LDEV1", but if you've been using Store-to-disk to store changes on ldev1 to "somewhere else" [private volume?], you should be able to recover them.

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *

ATOM RSS1 RSS2