Synology - BTRFS volume crashed & recovered. Why did it crash?

Solution 1:

I visited the WebGUI. There were no errors at all in the system other than the volume being offline.

  • The SMART status of each disk was checked and seemed OK
  • "Storage Manager" showed "Healthy"
  • "Storage Pool" showed "Healthy"

Only "volume" showed crashed.

I SSH'ed in and checked. mdadm --D /dev/md2 which is where my array was. It was showing State: Clean, Degraded

I checked dmesg, and found this:

[5638907.327288] ------------[ cut here ]------------
[5638907.332247] WARNING: CPU: 3 PID: 10234 at fs/btrfs/extent-tree.c:4207 btrfs_write_dirty_block_groups+0x365/0x390 [btrfs]()
[5638907.343601] BTRFS: Transaction aborted (error -2)
[5638907.343603] Modules linked in: nfsd exportfs rpcsec_gss_krb5 cifs udf isofs loop tcm_loop(O) iscsi_target_mod(O) target_core_ep(O) target_core_multi_file(O) target_core_file(O) target_core_iblock(O) target_core_mod(O) syno_extent_pool(PO) rodsp_ep(O) hid_generic usbhid hid usblp usb_storage denverton_synobios(PO) overlay exfat(O) btrfs synoacl_vfs(PO) hfsplus md4 hmac bnx2x(O) mdio mlx5_core(O) mlx4_en(O) mlx4_core(O) mlx_compat(O) qede(O) qed(O) atlantic(O) tn40xx(O) i40e(O) ixgbe(O) be2net(O) igb(O) i2c_algo_bit e1000e(O) vxlan ip6_udp_tunnel udp_tunnel fuse vfat fat crc32c_intel aesni_intel glue_helper lrw gf128mul ablk_helper arc4 cryptd ecryptfs sha256_generic ecb aes_x86_64 authenc des_generic ansi_cprng cts md5 cbc cpufreq_powersave cpufreq_performance acpi_cpufreq processor cpufreq_stats
[5638907.425092]  dm_snapshot dm_bufio crc_itu_t crc_ccitt quota_v2 quota_tree psnap p8022 llc sit tunnel4 ip_tunnel ipv6 zram sg etxhci_hcd xhci_pci xhci_hcd uhci_hcd ehci_pci ehci_hcd usbcore usb_common [last unloaded: denverton_synobios]
[5638907.448308] CPU: 3 PID: 10234 Comm: btrfs-transacti Tainted: P           O    4.4.59+ #25556
[5638907.457047] Hardware name: Synology Inc. RS2818RP+/Type2 - Board Product Name1, BIOS M.212 2019/11/01
[5638907.466571]  0000000000000000 ffff880068a0fc50 ffffffff812bf70d ffff880068a0fc98
[5638907.474851]  ffffffffa0939b8d ffff880068a0fc88 ffffffff8104b7cd ffff8801704f9e00
[5638907.483132]  ffff88003b616338 0000000000000001 00000000fffffffe ffff8801704f9f50
[5638907.491422] Call Trace:
[5638907.494179]  [<ffffffff812bf70d>] dump_stack+0x4d/0x70
[5638907.499623]  [<ffffffff8104b7cd>] warn_slowpath_common+0x7d/0xc0
[5638907.505932]  [<ffffffff8104b859>] warn_slowpath_fmt+0x49/0x50
[5638907.511996]  [<ffffffffa0892d45>] btrfs_write_dirty_block_groups+0x365/0x390 [btrfs]
[5638907.520056]  [<ffffffffa0932df8>] commit_cowonly_roots+0x230/0x2d1 [btrfs]
[5638907.527250]  [<ffffffffa08a90e8>] btrfs_commit_transaction+0x528/0xcb0 [btrfs]
[5638907.534793]  [<ffffffffa08a9905>] ? start_transaction+0x95/0x3d0 [btrfs]
[5638907.541810]  [<ffffffffa08a387c>] transaction_kthread+0x1ec/0x220 [btrfs]
[5638907.548915]  [<ffffffffa08a3690>] ? btrfs_cleanup_transaction+0x510/0x510 [btrfs]
[5638907.556701]  [<ffffffff810672a6>] kthread+0xc6/0xe0
[5638907.561883]  [<ffffffff810671e0>] ? kthread_create_on_node+0x180/0x180
[5638907.568717]  [<ffffffff81567abf>] ret_from_fork+0x3f/0x80
[5638907.574423]  [<ffffffff810671e0>] ? kthread_create_on_node+0x180/0x180
[5638907.581346] ---[ end trace 27185b26c2db1370 ]---
[5638907.586280] BTRFS: error (device md2) in btrfs_write_dirty_block_groups:4207: errno=-2 No such entry
[5638907.595721] BTRFS info (device md2): forced readonly
[5638907.600997] BTRFS warning (device md2): Skipping commit of aborted transaction.
[5638907.608618] BTRFS: error (device md2) in cleanup_transaction:2019: errno=-2 No such entry
[5638907.617108] BTRFS info (device md2): delayed_refs has NO entry

So the data was there, and the array was in read only. My research took me to this SuSe KB: https://www.suse.com/support/kb/doc/?id=000018769

I references the same error as I got BTRFS: Transaction aborted (error -2).

The article states

Good thing is, it's a WARNING, not a fatal error. WARNINGs like this one, e.g. regarding quota, typically are runtime only things that are fixed by BTRFS after the WARNING is issued. Not a bad problem.

Which was somewhat reassuring.

I ran a

syno_poweroff_task -d

In order to shutdown all the Synology services that may be accessing the volume. This stops the WebUI etc., but keeps SSH on.

I then did a

umount /volume1 

In order to stop I/O to the volume (although it was already in RO mode per the output of dmesg above. I then did a btrfsck on the md2 device. Output below

# btrfsck /dev/md2
Syno caseless feature on.
Checking filesystem on /dev/md2
UUID: 7a29febb-e9b5-4f77-afd7-4e1e10971340
checking extents
checking free space tree
checking fs roots
checking csums
checking root refs
found 30769182859264 bytes used err is 0
total csum bytes: 44052
total tree bytes: 50331648
total fs tree bytes: 24477696
total extent tree bytes: 23691264
btree space waste bytes: 1190925
file data blocks allocated: 30769149861888
referenced 30769131089920

And in dmesg:

[5644451.646580] BTRFS info (device md127): using free space tree
[5644451.652561] BTRFS info (device md127): has skinny extents
[5644459.827213] BTRFS info (device md127): checking UUID tree

Once that completed, I simply rebooted the system and everything came up as per normal.

I'm currently running a volume consistency check" on it which I believe is a mdadm "resync" that's been running for almost 24 hours now with no issues.

I guess the goal of making this post is to figure out if anyone with a bit more knowhow on btrfs has experienced this, and if they have any ideas what caused this?