Bug 60522 - btrfs_run_delayed_refs:2504: Object already exists (run_one_delayed_ref returned -17)
Summary: btrfs_run_delayed_refs:2504: Object already exists (run_one_delayed_ref retur...
Status: NEW
Alias: None
Product: File System
Classification: Unclassified
Component: btrfs (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: Josef Bacik
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2013-07-05 15:54 UTC by David F.
Modified: 2016-03-20 10:11 UTC (History)
2 users (show)

See Also:
Kernel Version: 3.8.13
Tree: Mainline
Regression: No


Attachments
relevant dmesg output (1.75 KB, text/x-log)
2013-07-05 15:54 UTC, David F.
Details

Description David F. 2013-07-05 15:54:08 UTC
Created attachment 106812 [details]
relevant dmesg output

Setup description:

Label: none  uuid: d7662fe5-04fc-4b4b-a6e0-4d34637d56d3
	Total devices 2 FS bytes used 188.91GB
	devid    1 size 500.00GB used 279.03GB path /dev/sda8
	devid    2 size 500.00GB used 279.01GB path /dev/sdb3

The devices are in raid1 mode and the volume is mounted at /home with noatime,compress=lzo,space_cache,inode_cache.

Bug description:
I ran into this bug a few days ago. My machine has been running for many, many hours with suspends to RAM in between. Suddenly the affected BTRFS volume became read-only. After a reboot (which required pushing the reset button because the volume couldn't be unmounted), the problem reoccured with virtually the same dmesg output (see attachment) within a few minutes.
I fsck'ed and scrubbed the volume in single user mode without errors.
After another reboot, same problem within minutes. For the time being I worked around the problem by copying relevant data to another partition and shut my machine down after work.
Yesterday I scrubbed it again in multiuser mode but without logging into a desktop session. Afterwards I resumed using the volume normally and it didn't fail until now.

I'm not involved with btrfs or kernel development but willing and able to poke around more if necessary.
Comment 1 David F. 2013-07-05 15:58:44 UTC
A remark to the attached dmesg log:
There was no message in the preceeding 100 seconds. In the following ~100 seconds were only messages about other components complaining about /home being unwritable.
Comment 2 Josef Bacik 2013-07-09 00:49:37 UTC
I believe I've fixed this already, could you try and reproduce on btrfs-next and see if you still see the problem?  If you can't build your own kernel I'm pretty sure the patch is in 3.9.  Thanks.
Comment 3 David F. 2013-07-09 13:49:45 UTC
Alright, I'll try to find the relevant change set, though I have no idea how to reproduce it. (I didn't find any reference to the same problem in the first place; the closest it got, was http://www.spinics.net/lists/linux-btrfs/msg17504.htm)

Note You need to log in before you can comment on or make changes to this bug.