Re: [PATCH v2 2/2] btrfs: fix missing superblock update in the device delete commit transaction

[Anand Jain <anand.jain@oracle.com>]

On 07/24/2018 07:01 PM, David Sterba wrote:
> On Tue, Jul 24, 2018 at 05:28:03PM +0800, Lu Fengqi wrote:
>>> I can't reproduce the issue. Do you reproduce consistently? and I
>>
>> Sorry I've taken so long to reply.
>>
>> There are four virtual machines with the different storage backend here.
>> And I can reproduce this issue consistently on the virtual machines (with
>> HDD or qcow2 file in HDD), however, I can't reproduce on the virtual machine
>> (with SSD or qcow2 file in SSD). So this may depend on the disk IO speed.
>>
>>> wonder if your workspace contains the latest changes from
>>> kdave/misc-next? Because last weeks kdave/misc had some issues.
>>> Can you please give a try?
>>
>> I used the latest kdave/misc-next branch on July 21, 2018, when I send
>> the previous mail. Just as David replied to Nikolay's thread, the current
>> latest kdave/misc-next still has the same problem.
> 
> I hit that on a physical machine with a mix of HDD and SSD drives under
> fstests. The VM tests did not hit the problem.


I couldn't reproduce this no matter what I try (physical machine, HDD
SDD,vbox, etc). But theoretically I have found what is going wrong.

This patch only let the problem in the btrfs_shrink_device() to
surface and it did not create it. The problem is due to
btrfs_shrink_device() did not call the commit transaction after the
device has been added to the fs_devices::resize_devices, which before
(this patch) the btrfs_rm_dev_item() did the commit transaction for the
btrfs_shrink_device().

I am sending a patch to fix. Thanks;

Anand





> --
> To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>