iSCSI LIO-Target process crashes on load leading to file system corruption

[tseln]

Hello everybody, 

the old page allocation error again with LIO-target.  Firmware is 6.10.3, still no NUMA support in the kernel, therefore the antique 

sysctl -w vm.zone_reclaim_mode = 1 cannot be executed.

The problem can be reliably triggered by starting a balance on the btrfs the LUN (logical drive) is formatted with. I guess it's aggravated by the 10G link, write load is certainly heavy.

The first complaint about this problem is from 2016 in this forum. Is there still no solution other then throwing ReadyNASOS away and switch over to another linux/FreeBSD?

Any ideas how to fix this?

Tseln

iscsi_trx: page allocation failure: order:4, mode:0x24040c0
[278412.450255] CPU: 0 PID: 19316 Comm: iscsi_trx Tainted: P O 4.4.190.x86_64.1 #1
[278412.450257] Hardware name: NETGEAR ReadyNAS 524X/ReadyNAS 524X, BIOS RN524v22 11/29/2016
[278412.450258] 0000000000000000 ffff88015f1d3b98 ffffffff8836360d 00000000024040c0
[278412.450262] 0000000000000001 ffff88015f1d3c20 ffffffff880e2749 0000000400000040
[278412.450264] 0000000088d80080 ffff880054911780 0000000000000004 0000000000000040
[278412.450267] Call Trace:
[278412.450274] [<ffffffff8836360d>] dump_stack+0x4d/0x63
[278412.450277] [<ffffffff880e2749>] warn_alloc_failed+0x10c/0x121
[278412.450280] [<ffffffff880e4d86>] __alloc_pages_nodemask+0x7f8/0x82f
[278412.450282] [<ffffffff880e4f8f>] alloc_kmem_pages+0x1d/0x83
[278412.450286] [<ffffffff880f6345>] kmalloc_order+0x13/0x3a
[278412.450290] [<ffffffff881112e3>] __kmalloc+0x2f/0x12b
[278412.450294] [<ffffffff885760db>] target_alloc_sgl+0x4a/0x127
[278412.450298] [<ffffffff888e6121>] ? _raw_spin_unlock_bh+0x15/0x17
[278412.450300] [<ffffffff885762f8>] transport_generic_new_cmd+0x140/0x1f4
[278412.450303] [<ffffffff8858fcd1>] iscsit_process_scsi_cmd+0x90/0xa5
[278412.450305] [<ffffffff8859286c>] iscsi_target_rx_thread+0x332/0xca1
[278412.450307] [<ffffffff888e39dc>] ? __schedule+0x3f7/0x4f2
[278412.450309] [<ffffffff888e39e8>] ? __schedule+0x403/0x4f2
[278412.450311] [<ffffffff8859253a>] ? iscsi_target_tx_thread+0x1c0/0x1c0
[278412.450315] [<ffffffff88079cbc>] kthread+0xdc/0xe4
[278412.450317] [<ffffffff888e39e8>] ? __schedule+0x403/0x4f2
[278412.450320] [<ffffffff888e0000>] ? soft_show+0x6/0x27
[278412.450323] [<ffffffff88079be0>] ? kthread_worker_fn+0x129/0x129
[278412.450325] [<ffffffff888e6845>] ret_from_fork+0x55/0x80
[278412.450328] [<ffffffff88079be0>] ? kthread_worker_fn+0x129/0x129
[278412.450329] Mem-Info:
[278412.450333] active_anon:13631 inactive_anon:17385 isolated_anon:0
active_file:420038 inactive_file:442451 isolated_file:32
unevictable:80 dirty:20562 writeback:0 unstable:0
slab_reclaimable:18077 slab_unreclaimable:9267
mapped:3246 shmem:3327 pagetables:2171 bounce:0
free:30907 free_pcp:173 free_cma:0