Simon Riggs wrote:
On Fri, 2008-01-25 at 17:56 +0100, Zdenek Kotala wrote:
Regarding to Robert Mach's work during Google SOC on data integrity
check. I would like to improve storage module and implement some
Robert's code into the core.
I would like to make following modification:
1) Add ReadBuffer_noerror (recommend me better name) function which will
accept damaged page without Error. This page will be marked as corrupted
and when ReadBuffer will touch this page then it will be handled in
standard way.
This is important for check and repair functions to process all table
without interruption.
We shouldn't let duff data into shared buffers at all.
As Tom mentioned before. I agree, it could cause a lot of problems.
I think you could mix the two methods of reading buffers
- start a subtransaction
- read blocks into shared buffers
- if error, then re-read block into private memory and examine
- carry on thru table in a new subtransaction
It seems like good idea.
OK with other points, except I don't want a new command. Let's do it as
a function that can accept block ranges to check, not just whole tables.
e.g. pg_check_blocks(17, 43) would check blocks 17 -> 43
It makes sense. I think following function should cover all cases:
pg_check_blocks() - all db
pg_check_blocks(relno) - all relation
pg_check_blocks(relno, start, stop) - selected interval
pg_check_blocks(relno, array of blocks) - selected blocks
Zdenek
---------------------------(end of broadcast)---------------------------
TIP 5: don't forget to increase your free space map settings
