| To: [email protected]
Cc: [email protected]
Subject: Re: question
In-Reply-To: Your message of "Tue, 28 Jan 97 11:14:58 EST."
<[email protected]>
Date: Tue, 28 Jan 97 17:30:03 -0500
>
> Do you have a better way of fixing a corrupted database.
> Jeff Kent from Trident says he occasionally runs into the follow problem afte
> r
> a hardware swap and a occasional reboot. During the bootup the system
> keeps cycling (ie) trying to boot multiusermode, but never gets up. They
> interupt the boot and can boot to single usermode. Jeff says that the
> MACILDBASE file gets corrupted. Lee gave the following instructions to
> give to Trident. He said you may have a better way.
>
>
> I think the following can be done:
>
> o boot single user
> o run dbck with the proper args to
> check/repair the database
>
The above is the correct way to fix a corrupted MACILBDBASE.
(I actually filed a QAR on this many years ago -- the system
boot procedure should run dbck on the database when the system
starts up -- and if it fails, boot to single-user -- not just
sit there in an infinite loop -- would be a simple matter to
do this, just modify the start-up shell scripts).
but I am concerned about why this is happening.
If it is a question of questionable blocks on the disk,
i.e., the MACILBDBASE just happens to be sitting on some
flakey patches on the disk,
then then next time it happens, fix the database then
rename it to something else (do not remove it), and then
copy it to another file with the real name and check that.
This will keep the "bad" version located at the "bad" blocks
and new area of the disk will be used for the replacement.
And the "bad" blocks won't get re-used for any other files.
But if there is something going on with 'hardware swap'
then I'd like to know.
What is the "hardware swap" procedure -- are they shutting
down the systems properly? Maybe shutdown is not umounting
the root file system and closing up daemons properly...
Also, is this V2.1, V3.1A or T4.0?
regards,
BAT
|