On Wed, Aug 30, 2017 at 03:56:05PM +0200, Christian Theune wrote:

Hi,

just got it again on a different call path, maybe that helps:

[ 1070.136303] Oops: 0000 [#1] SMP
[ 1070.142577] Modules linked in: nf_log_ipv4 nf_conntrack_ipv4 nf_defrag_ipv4 nf_log_ipv6 nf_log_common xt_LOG xt_limit nf_conntrack_ipv6 nf_defrag_ipv6 xt_conntrack sch_fq x86_pkg_temp_thermal kvm_intel kvm irqbypass nvme crc32c_intel ixgbe nvme_core mdio acpi_cpufreq nbd nf_conntrack_ftp nf_conntrack dm_zero dm_thin_pool dm_persistent_data dm_bio_prison dm_round_robin dm_multipath xts aesni_intel glue_helper lrw ablk_helper cryptd aes_x86_64 fuse dm_snapshot dm_bufio dm_crypt dm_mirror dm_region_hash dm_log
[ 1070.233784] CPU: 19 PID: 7460 Comm: ceph-osd Not tainted 4.9.43 #1
[ 1070.246124] Hardware name: Thomas-Krenn.AG X9DR3-F/X9DR3-F, BIOS 3.0a 07/31/2013
[ 1070.260895] task: ffff8810517d0000 task.stack: ffffc9002abec000
[ 1070.272710] RIP: 0010:[<ffffffff81312320>]  [<ffffffff81312320>] xfs_da3_node_read+0x30/0xb0
[ 1070.289592] RSP: 0018:ffffc9002abefd28  EFLAGS: 00010286
[ 1070.300199] RAX: 0000000000000000 RBX: ffff88104d859a48 RCX: 0000000000000001
[ 1070.314447] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffffc9002abefce0
[ 1070.328694] RBP: ffffc9002abefd48 R08: 0000000066656566 R09: ffffc9002abefbc0
[ 1070.342942] R10: fffffffffffffffe R11: 0000000000000001 R12: ffffc9002abefd78
[ 1070.357191] R13: ffff88066b430780 R14: 0000000000000005 R15: 0000000066656566
[ 1070.371436] FS:  00007fe511bfc700(0000) GS:ffff88107fbc0000(0000) knlGS:0000000000000000
[ 1070.387590] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 1070.399066] CR2: 00000000000000a0 CR3: 0000000f14d50000 CR4: 00000000001406e0
[ 1070.413311] Stack:
[ 1070.417332]  ffffffff81a44fe0 ffffc9002abefd48 ffffc9002abefdd0 0000000000000005
[ 1070.432239]  ffffc9002abefdb8 ffffffff81337404 0000000200000008 ffff8809b5cab040
[ 1070.447144]  000000005e94ce38 ffff880c25e1c600 0000000000000000 0000000000000000
[ 1070.462051] Call Trace:
[ 1070.466949]  [<ffffffff81337404>] xfs_attr3_node_inactive+0x174/0x210
[ 1070.479802]  [<ffffffff813376da>] xfs_attr_inactive+0x23a/0x250
[ 1070.491625]  [<ffffffff81350a4b>] xfs_inactive+0x7b/0x110
[ 1070.502403]  [<ffffffff81359344>] xfs_fs_destroy_inode+0xa4/0x210
[ 1070.514573]  [<ffffffff811c46cb>] destroy_inode+0x3b/0x60
[ 1070.525352]  [<ffffffff811c4819>] evict+0x129/0x190
[ 1070.535093]  [<ffffffff811c4c4a>] iput+0x19a/0x200
[ 1070.544660]  [<ffffffff811b9129>] do_unlinkat+0x129/0x2d0
[ 1070.555445]  [<ffffffff811b9d26>] SyS_unlink+0x16/0x20
[ 1070.565706]  [<ffffffff81885260>] entry_SYSCALL_64_fastpath+0x13/0x94

This looks like the same call stack as last time.

Is this with a patched 4.9.43 kernel, or just vanilla?

--D

[ 1070.578562] Code: 55 48 89 e5 41 54 53 4d 89 c4 48 89 fb 48 83 ec 10 48 c7 04 24 e0 4f a4 81 e8 fd fe ff ff 85 c0 75 46 48 85 db 74 41 49 8b 34 24 <48> 8b 96 a0 00 00 00 0f b7 52 08 66 c1 c2 08 66 81 fa be 3e 74
[ 1070.618459] RIP  [<ffffffff81312320>] xfs_da3_node_read+0x30/0xb0
[ 1070.630663]  RSP <ffffc9002abefd28>
[ 1070.637630] CR2: 00000000000000a0
[ 1070.644858] ---[ end trace bc2d3667eef00f69 ]—

As of now the system doesn’t have the same following issues and the other FS’s are still functioning. I’ll run xfs_repair later today on all filesystems for good measure.

Christian

quoted

On Aug 28, 2017, at 9:00 PM, Christian Theune [off-list ref] wrote:

Hi,

quoted

On Aug 28, 2017, at 7:42 PM, Darrick J. Wong [off-list ref] wrote:

On Mon, Aug 28, 2017 at 07:23:19PM +0200, Christian Theune wrote:

quoted

Hi,

we stumbled over this today as a host rebooted with an unrelated (iommu)
kernel crash and got completely stuck after this:

I’m currently running xfs_repair on all disks and will then see whether this
will resolve, still I guess you want to know about it. Kernel is 4.9.43
vanilla. Let me know if you need more data.

Does commit cd87d8679201 ("xfs: don't crash on unexpected holes in dir/attr
btrees") fix this problem?  It'll be in 4.13, maybe someone can backport it
to 4.9?

Thanks for the suggestion. I’ll keep that in mind in case I see this again.

quoted

(Assuming you can get it to reproduce reliably?)

I have only seen it once today and hopefully won’t see it again. We have had some storage servers that run multiple SSD and HDD disks (for Ceph) crash multiple times a week lastly due to the IOMMU issues that resulted in hardware watchdog reboots, so I guess those xfs' did have quite some noise in it.

Not sure I can do anything to reproduce it at all. *fingers crossed*

Christian

--
Christian Theune · ct@flyingcircus.io · +49 345 219401 0
Flying Circus Internet Operations GmbH · http://flyingcircus.io
Forsterstraße 29 · 06112 Halle (Saale) · Deutschland
HR Stendal HRB 21169 · Geschäftsführer: Christian Theune, Christian Zagrodnick

Liebe Grüße,
Christian Theune

--
Christian Theune · ct@flyingcircus.io · +49 345 219401 0
Flying Circus Internet Operations GmbH · http://flyingcircus.io
Forsterstraße 29 · 06112 Halle (Saale) · Deutschland
HR Stendal HRB 21169 · Geschäftsführer: Christian Theune, Christian Zagrodnick