Re: [HACKERS] Online enabling of page level checksums - Mailing list pgsql-hackers

From Tom Lane
Subject Re: [HACKERS] Online enabling of page level checksums
Date
Msg-id 16358.1485221664@sss.pgh.pa.us
Whole thread Raw
In response to Re: [HACKERS] Online enabling of page level checksums  (Jim Nasby <Jim.Nasby@BlueTreble.com>)
List pgsql-hackers
Jim Nasby <Jim.Nasby@BlueTreble.com> writes:
> For a first pass, I think it's acceptable for autovac and vac to notice 
> if a relation needs checksums computed and ignore the VM in that case 
> (make sure it's ignoring all frozen bits too).

> Likewise, for right now I think it's OK to force users that are enabling 
> this to manually connect to datallowcon=false and run vacuum.

I think it's a *complete* mistake to overload vacuum with this
responsibility.  Build a separate tool with a separate scheduling policy.

As one reason why not: vacuum doesn't generally attempt to scan indexes
sequentially at all.  Some of the index AMs might happen to do that, but
touching every page of an index is nowhere in vacuum's charter.  Nor is
there a good reason for index AMs to be involved in the job of placing
checksums, but they'd all have to know about this if you insist on going
through vacuum.
        regards, tom lane



pgsql-hackers by date:

Previous
From: Jim Nasby
Date:
Subject: Re: [HACKERS] Checksums by default?
Next
From: Peter Geoghegan
Date:
Subject: Re: [HACKERS] Checksums by default?