Re: [bug] scrub on low disk space leads to Transaction aborted

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

On 11/1/21 8:38 am, Qu Wenruo wrote:
> 
> 
> On 2021/1/9 下午4:53, Anand Jain wrote:
>>
>>   I ran scrub when disk space was low on a btrfs with some raid1 csum
>>   errors, on a system with kernel 5.11.0-rc2+. It lead to transaction
>>   aborted and rdonly FS.
>>
> 
> Would you please provide `btrfs fi usage` output?
> 
> There is a bug in recent kernel that global rsv is not increased
> properly so that we have a much higher chance to exhaust metadata space.
> 

Now I have made some space in the filesystem. So the usage below may not 
be relevant at the time of the problem.

# btrfs fi usage /Volumes/
Overall:
     Device size:		 154.97GiB
     Device allocated:		 153.12GiB
     Device unallocated:		   1.85GiB
     Device missing:		  76.56GiB
     Used:			 147.63GiB
     Free (estimated):		   3.35GiB	(min: 3.35GiB)
     Data ratio:			      2.00
     Metadata ratio:		      2.00
     Global reserve:		 181.56MiB	(used: 0.00B)

Data,RAID1: Size:75.53GiB, Used:73.11GiB (96.79%)
    /dev/sda6	  75.53GiB
    /dev/sdb2	  75.53GiB

Metadata,RAID1: Size:1.00GiB, Used:724.14MiB (70.72%)
    /dev/sda6	   1.00GiB
    /dev/sdb2	   1.00GiB

System,RAID1: Size:32.00MiB, Used:16.00KiB (0.05%)
    /dev/sda6	  32.00MiB
    /dev/sdb2	  32.00MiB

Unallocated:
    /dev/sda6	   1.85GiB
    /dev/sdb2	   1.00MiB



> Thanks,
> Qu
> 
>> [ 2365.314375] ------------[ cut here ]------------
>> [ 2365.314385] BTRFS: Transaction aborted (error -28)
>> [ 2365.314470] WARNING: CPU: 2 PID: 4672 at fs/btrfs/inode.c:2823
>> btrfs_finish_ordered_io+0x8e6/0x950 [btrfs]
>> <snap>
>> [ 2365.314921] CPU: 2 PID: 4672 Comm: kworker/u8:11 Not tainted
>> 5.11.0-rc2+ #3
>> [ 2365.314933] Hardware name: LENOVO 20FAS6PN00/20FAS6PN00, BIOS
>> N1CET47W (1.15 ) 08/08/2016
>> [ 2365.314939] Workqueue: btrfs-endio-write btrfs_work_helper [btrfs]
>> [ 2365.315093] RIP: 0010:btrfs_finish_ordered_io+0x8e6/0x950 [btrfs]
>> <snap>
>> [ 2365.315290] Call Trace:
>> [ 2365.315309] BTRFS warning (device sda6):
>> btrfs_finish_ordered_io:2823: Aborting unused transaction(No space left).
>> [ 2365.315305]  finish_ordered_fn+0x15/0x20 [btrfs]
>> [ 2365.315532]  btrfs_work_helper+0xcc/0x2f0 [btrfs]
>> [ 2365.315763]  process_one_work+0x1ee/0x390
>> [ 2365.315781]  worker_thread+0x50/0x3d0
>> [ 2365.315797]  kthread+0x114/0x150
>> [ 2365.315811]  ? process_one_work+0x390/0x390
>> [ 2365.315824]  ? kthread_park+0x90/0x90
>> [ 2365.315836]  ret_from_fork+0x22/0x30
>> [ 2365.315861] ---[ end trace b0e289a2c3d424e8 ]---
>> [ 2365.315871] BTRFS warning (device sda6):
>> btrfs_finish_ordered_io:2823: Aborting unused transaction(No space left).
>> [ 2365.316713] BTRFS warning (device sda6):
>> btrfs_finish_ordered_io:2823: Aborting unused transaction(No space left).
>> [ 2365.316796] BTRFS warning (device sda6): Skipping commit of aborted
>> transaction.
>> [ 2365.316809] BTRFS: error (device sda6) in cleanup_transaction:1938:
>> errno=-30 Readonly filesystem
>> [ 2365.316824] BTRFS info (device sda6): forced readonly
>> [ 2365.359195] BTRFS info (device sda6): scrub: not finished on devid 2
>> with status: -125