Re: [CORRUPTION FILESYSTEM] Corrupted and unrecoverable file system during the snapshot receive

[bepi@adria.it]

Hi.

I will insert ' btrfs check ' after each ' receive ' in my script.

I will test again my hardware.
But is not very likely that 2 computers, 3 HDD, 3 partitions, all have issue.

I think that the problem is a concomitance of operations, a race condition, a
random conditions.
I'll try to create a test case.


P.S. For find the problem may need to insert tools as ' coredumper ' and  '
sanitize ' in ' btrfs ', detect in realtime the ' extent ' corruption, and log
detection.


Thank you.

Gdb



Xin Zhou <xin.zhou@gmx.com>:

> Hi,
> 
> The system seems running some customized scripts continuously backup data
> from a NVME drive to HDDs.
> If the 3 HDDs backup storage are same in btrfs config, and the there is a bug
> in btrfs code,
> they all suppose to fail after the same operation sequence.
> 
> Otherwise, probably one of the HDDs might have issue, or there is a bug in
> layer below btrfs.
> 
> For the customize script, it might be helpful to check the file system
> consistency after each transfer.
> That might be useful to figure out which step generates a corruption, and if
> there is error propagations.
> 
> Regards,
> Xin
>  
>  
> 
> Sent: Monday, December 19, 2016 at 10:55 AM
> From: "Giuseppe Della Bianca" <bepi@adria.it>
> To: "Xin Zhou" <xin.zhou@gmx.com>
> Cc: linux-btrfs@vger.kernel.org
> Subject: Re: [CORRUPTION FILESYSTEM] Corrupted and unrecoverable file system
> during the snapshot receive
> a concrete example
> 
> 
> SNAPSHOT
> 
> /dev/nvme0n1p2 on /tmp/tmp.X3vU6dLLVI type btrfs
> (rw,relatime,ssd,space_cache,subvolid=5,subvol=/)
> 
> btrfsManage SNAPSHOT /
> 
> (2016-12-19 19:44:00) Start btrfsManage
> . . . Start managing SNAPSHOT ' / ' filesystem ' root ' snapshot
> 
> In ' btrfssnapshot ' latest source snapshot ' root-2016-12-18_15:10:01.40 '
> . . . date ' 2016-12-18_15:10:01 ' number ' 40 '
> 
> Creation ' root-2016-12-19_19:44:00.part ' snapshot from ' root ' subvolume
> . . . Create a readonly snapshot of '/tmp/tmp.X3vU6dLLVI/root' in
> '/tmp/tmp.X3vU6dLLVI/btrfssnapshot/root/root-2016-12-19_19:44:00.part'
> 
> Renaming ' root-2016-12-19_19:44:00.part ' into ' root-2016-12-19_19:44:00.41
> ' snapshot
> 
> Source snapshot list of ' root ' subvolume
> . . . btrfssnapshot/root/root-2016-08-28-12-35-01.1
> ]zac[
> . . . btrfssnapshot/root/root-2016-12-19_19:44:00.41
> 
> (2016-12-19 19:44:05) End btrfsManage
> . . . End managing SNAPSHOT ' / ' filesystem ' root ' snapshot
> CORRECTLY
> 
> 
> 
> SEND e RECEIVE
> 
> /dev/nvme0n1p2 on /tmp/tmp.o78czE0Bo6 type btrfs
> (rw,relatime,ssd,space_cache,subvolid=5,subvol=/)
> /dev/sda2 on /tmp/tmp.XcwqQCKq09 type btrfs
> (rw,relatime,space_cache,subvolid=5,subvol=/)
> 
> btrfsManage SEND / /dev/sda2
> 
> (2016-12-19 19:47:24) Start btrfsManage
> . . . Start managing SEND ' / ' filesystem ' root ' snapshot in ' /dev/sda2
> '
> 
> Sending ' root-2016-12-19_19:44:00.41 ' source snapshot to ' btrfsreceive '
> subvolume
> . . . btrfs send -p
> /tmp/tmp.o78czE0Bo6/btrfssnapshot/root/root-2016-12-18_15:10:01.40
> /tmp/tmp.o78czE0Bo6/btrfssnapshot/root/root-2016-12-19_19:44:00.41 | btrfs
> receive /tmp/tmp.XcwqQCKq09/btrfsreceive/root/.part/
> . . . At subvol
> /tmp/tmp.o78czE0Bo6/btrfssnapshot/root/root-2016-12-19_19:44:00.41
> . . . At snapshot root-2016-12-19_19:44:00.41
> 
> Creation ' root-2016-12-19_19:44:00.41 ' snapshot from '
> .part/root-2016-12-19_19:44:00.41 ' subvolume
> . . . Create a readonly snapshot of
> '/tmp/tmp.XcwqQCKq09/btrfsreceive/root/.part/root-2016-12-19_19:44:00.41' in
> '/tmp/tmp.XcwqQCKq09/btrfsreceive/root/root-2016-12-19_19:44:00.41'
> . . . Delete subvolume (commit):
> '/tmp/tmp.XcwqQCKq09/btrfsreceive/root/.part/root-2016-12-19_19:44:00.41'
> 
> Snapshot list in ' /dev/sda2 ' device
> . . . btrfsreceive/data_backup/data_backup-2016-12-17_12:07:00.1
> . . . btrfsreceive/data_storage/data_storage-2016-12-10_17:05:51.1
> . . . btrfsreceive/root/root-2016-08-28-12-35-01.1
> ]zac[
> . . . btrfsreceive/root/root-2016-12-19_19:44:00.41
> 
> (2016-12-19 19:48:37) End btrfsManage
> . . . End managing SEND ' / ' filesystem ' root ' snapshot in ' /dev/sda2 '
> CORRECTLY
> 
> 
> 
> > Hi Giuseppe,
> >
> > Would you like to tell some details about:
> > 1. the XYZ snapshot was taken from which subvolume
> > 2. where the base (initial) snapshot is stored
> > 3. The 3 partitions receives the same snapshot, are they in the same btrfs
> > configuration and subvol structure?
> >
> > Also, would you send the link reports "two files unreadable error" post
> > mentioned in step 2? Hope can see the message and figure out if the issue
> > first comes from sender or receiver side.
> >
> > Thanks,
> > Xin
> >
> >
>  
> 




----------------------------------------------------
This mail has been sent using Alpikom webmail system
http://www.alpikom.it