On Fri, Mar 01, 2013 at 05:30:57PM +0100, David Sterba wrote:
> The mail did not make it to the list on Monday, anyway, now I have a
> few testing results to share.
>
> The umount responsiveness is much improved, it took a few seconds when
> there were many roots to clean scheduled.
>
> With
>
> $ echo 'func btrfs_drop_snapshot +pf' >
> /sys/debug/kernel/dynamic_debug/control
> $ echo 'func btrfs_clean_one_deleted_snapshot +pf' >
> /sys/debug/kernel/dynamic_debug/control
>
> one can watch how it proceeds.
>
> After umount and and mount, the cleaning process starts again, after 30
> seconds when transaction commit triggers.
>
> Next test was to let it run with balance (just 'fi balance', no other
> options). Worked well, but I've hit an oops from balance cancel command that
> was triggered sometime during the balance. According to the log it happened
> right after the balance finished. CCing Ilya, I'm not sure if this is somehow
> induced by the patch.
>
> 3407 BUG_ON(fs_info->balance_ctl ||
> atomic_read(&fs_info->balance_running));
Were you by any chance running this on top of a for-linus that included
a raid56 merge commit? If so, this is result of a raid56 mismerge --
fs_info->balance_ctl is supposed to be cleared in __cancel_balance(), a
call to which was accidentally nuked.
Thanks,
Ilya
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
