Comment 21 for bug 1891773

Just happened for the second time in 46 hours (and it never happened before):

kernel: [167836.884337] INFO: task kcompactd0:63 blocked for more than 120 seconds.
kernel: [167836.887341] Not tainted 4.15.0-128-generic #131-Ubuntu
kernel: [167836.889880] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
kernel: [167836.893754] kcompactd0 D 0 63 2 0x80000000
kernel: [167836.893760] Call Trace:
kernel: [167836.894017] __schedule+0x24e/0x880
kernel: [167836.894031] schedule+0x2c/0x80
kernel: [167836.894034] io_schedule+0x16/0x40
kernel: [167836.894160] __lock_page+0xff/0x140
kernel: [167836.894197] ? page_cache_tree_insert+0xe0/0xe0
kernel: [167836.894246] migrate_pages+0x91f/0xb80
kernel: [167836.894269] ? __ClearPageMovable+0x10/0x10
kernel: [167836.894272] ? isolate_freepages_block+0x3b0/0x3b0
kernel: [167836.894276] compact_zone+0x681/0x950
kernel: [167836.894279] kcompactd_do_work+0xfe/0x2a0
kernel: [167836.894282] ? __switch_to_asm+0x35/0x70
kernel: [167836.894284] ? __switch_to_asm+0x41/0x70
kernel: [167836.894288] kcompactd+0x86/0x1c0
kernel: [167836.894292] ? kcompactd+0x86/0x1c0
kernel: [167836.894395] ? wait_woken+0x80/0x80
kernel: [167836.894445] kthread+0x121/0x140
kernel: [167836.894449] ? kcompactd_do_work+0x2a0/0x2a0
kernel: [167836.894452] ? kthread_create_worker_on_cpu+0x70/0x70
kernel: [167836.894454] ret_from_fork+0x35/0x40

Message appears every 2 minutes for 18 minutes (then I guess it stops reporting) ; after that the disk are blocked, then NFS fails, then the whole service I'm running fails.

It happened twice with kernel 4.15.0-128-generic.

I updated to 4.15.0-137-generic while service was down, let's see how long it will work.
HWE kernel is the next step.