Re: [PATCH] Btrfs: scrub: batch rebuild for raid56

[David Sterba <dsterba@suse.cz>]

On Wed, Mar 07, 2018 at 05:22:08PM +0200, Nikolay Borisov wrote:
> 
> 
> On  7.03.2018 16:43, David Sterba wrote:
> > On Tue, Mar 06, 2018 at 11:22:21AM -0700, Liu Bo wrote:
> >> On Tue, Mar 06, 2018 at 11:47:47AM +0100, David Sterba wrote:
> >>> On Fri, Mar 02, 2018 at 04:10:37PM -0700, Liu Bo wrote:
> >>>> In case of raid56, writes and rebuilds always take BTRFS_STRIPE_LEN(64K)
> >>>> as unit, however, scrub_extent() sets blocksize as unit, so rebuild
> >>>> process may be triggered on every block on a same stripe.
> >>>>
> >>>> A typical example would be that when we're replacing a disappeared disk,
> >>>> all reads on the disks get -EIO, every block (size is 4K if blocksize is
> >>>> 4K) would go thru these,
> >>>>
> >>>> scrub_handle_errored_block
> >>>>   scrub_recheck_block # re-read pages one by one
> >>>>   scrub_recheck_block # rebuild by calling raid56_parity_recover()
> >>>>                         page by page
> >>>>
> >>>> Although with raid56 stripe cache most of reads during rebuild can be
> >>>> avoided, the parity recover calculation(xor or raid6 algorithms) needs to
> >>>> be done $(BTRFS_STRIPE_LEN / blocksize) times.
> >>>>
> >>>> This makes it less stupid by doing raid56 scrub/replace on stripe length.
> >>>
> >>> missing s-o-b
> >>
> >> I'm surprised that checkpatch.pl didn't complain.
> 
> I have written a python script that can scrape the mailing list and run
> checkpatch (and any other software deemed appropriate) on posted patches
> and reply back with results. However, I haven't really activated it, I
> guess if people think there is merit in it I could hook it up to the
> mailing list :)

If we and checkpatch agree on the issues to report then yes, but I think
it would be confusing when we'd have to tell people to ignore some of
the warnings. I think we'd need to tune checkpatch for our needs, I
would not mind testing it locally. I'd start with catching typos in the
changelogs and comments, that's something that happens all the time and
can be automated. Next, implement AI that will try to understand the
changelog and tell people what's missing.