Re: pgcheck - data integrity check - Mailing list pgsql-hackers

From Rober Mach
Subject Re: pgcheck - data integrity check
Date
Msg-id 46BE69E3.5050709@gmail.com
Whole thread Raw
In response to Re: pgcheck - data integrity check  (Josh Berkus <josh@agliodbs.com>)
List pgsql-hackers

Josh Berkus wrote:
> Robert,
>
>   
>> I am working on a data integrity check tool (pgcheck).
>> I would like to discuss the following issues:
>>     
>
> I'm a little confused.  I assumed that your project would check the pages of a 
> *shut-down* database or one in recovery (single-user) mode as part of 
> recovery after a crash or HW failure.  We don't really want users running the 
> database in normal mode if there's any significant chance of data-page 
> corruption.
>
> Can you back up and explain what you're trying to accomplish with this tool?
>
>   
So far I was taking it just as a tool for validating the data in a
database. A tool, which would be runned from psql and which would tell
the user, whether the data is ok or not, eventually if there is any
corrupted data, tell where the problem is... Just for checking the data
integrity I thing this is possible way.

It will be probably different in case of the data recovery tool and that
is what I am trying to get opinions on..  Maybe one possible way (but
probably not very convenient) is to run the check from psql to find out
where the problem is and then run recovery tool on the corrupted files
or use the "sigle-user" mode for recovery with the database being turned
off. The integrity check tool could be also implemented to be runned on
a *shut-down* database (for case of database crash), but I thing that
being able to run it under *normal* conditions should be helpful as well.

Robert


pgsql-hackers by date:

Previous
From: Andrew Dunstan
Date:
Subject: Re: Wrote a connect-by feature
Next
From: Tom Lane
Date:
Subject: Re: proper way to fix information_schema.key_column_usage view