kernel: “echo 0 / proc / sys / kernel / hung_task_timeout_secs”

1

Eu recebi esses erros no syslog. Alguma ideia do que isso significa? O sistema está rodando no Ubuntu 12.04, kernel: 3.8.0-35-generic #52~precise1

Parece-me falha ao tentar gravar no disco ...

 [151850.317166] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
 [151850.318428]  ffff8803640f5be8 0000000000000082 0000000034c733da ffff88040e233ec0
 [151850.318444]  ffff880408149740 ffff880403c6c5c0 0000000000000001 ffff8803edae80a8
 [151850.318468]  [<ffffffff816f41c9>] schedule+0x29/0x70
 [151850.318481]  [<ffffffff816f30b7>] __mutex_lock_slowpath+0xd7/0x150
 [151850.318497]  [<ffffffff816f2cca>] mutex_lock+0x2a/0x50
 [151850.318510]  [<ffffffff811aa006>] path_lookupat+0x236/0x7a0
 [151850.318523]  [<ffffffff811aade0>] ? getname_flags.part.31+0x30/0x150
 [151850.318537]  [<ffffffff811aaf6e>] ? getname_flags+0x6e/0x80
 [151850.318552]  [<ffffffff813141f4>] ? apparmor_inode_getattr+0x54/0x60
 [151850.318565]  [<ffffffff811bb929>] ? mntput_no_expire+0x49/0x160
 [151850.318578]  [<ffffffff811ab711>] user_path_at+0x11/0x20
 [151850.318588]  [<ffffffff811a0c3e>] vfs_lstat+0x1e/0x20
 [151850.318602]  [<ffffffff816fdf5d>] system_call_fastpath+0x1a/0x1f
 [151850.319634] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
 [151850.376552]  [<ffffffff816f41c9>] schedule+0x29/0x70
 [151850.376565]  [<ffffffff816f30b7>] __mutex_lock_slowpath+0xd7/0x150
 [151850.376579]  [<ffffffff816f2cca>] mutex_lock+0x2a/0x50
 [151850.376590]  [<ffffffff811aa006>] path_lookupat+0x236/0x7a0
 [151850.376602]  [<ffffffff811aade0>] ? getname_flags.part.31+0x30/0x150
 [151850.376616]  [<ffffffff811aaf6e>] ? getname_flags+0x6e/0x80
 [151850.376680]  [<ffffffff813141f4>] ? apparmor_inode_getattr+0x54/0x60
 [151850.376691]  [<ffffffff811bb929>] ? mntput_no_expire+0x49/0x160
 [151850.376704]  [<ffffffff811ab711>] user_path_at+0x11/0x20
 [151850.406554] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
 [151850.465132] sftp-server     D ffff880403bfdf60     0 15199      1 0x00000000
 [151850.465158]  ffff8803d2c87fd8 ffff8803d2c87fd8 ffff8803d2c87fd8 0000000000013ec0
 [151850.465173]  ffffffff81c15440 ffff88040583ae80 0000000000000001 ffff8803edae80a8
 [151850.465263]  [<ffffffff811aa006>] path_lookupat+0x236/0x7a0
 [151850.465269]  [<ffffffff81186441>] ? kmem_cache_alloc+0x31/0x140
 [151850.465275]  [<ffffffff811aade0>] ? getname_flags.part.31+0x30/0x150
 [151850.465283]  [<ffffffff811aa5a4>] filename_lookup+0x34/0xc0
 [151850.465289]  [<ffffffff811aaf6e>] ? getname_flags+0x6e/0x80
 [151850.465296]  [<ffffffff811ab6b9>] user_path_at_empty+0x59/0xa0
 [151850.465303]  [<ffffffff813141f4>] ? apparmor_inode_getattr+0x54/0x60
 [151850.465309]  [<ffffffff81087e9a>] ? lg_local_unlock+0x1a/0x20
 [151850.465315]  [<ffffffff811bb929>] ? mntput_no_expire+0x49/0x160
 [151850.465320]  [<ffffffff811a0946>] ? cp_new_stat+0x116/0x130
 [151850.465327]  [<ffffffff811ab711>] user_path_at+0x11/0x20
 [151850.465332]  [<ffffffff811a0bc1>] vfs_fstatat+0x51/0xb0
 [151850.465337]  [<ffffffff811a0c3e>] vfs_lstat+0x1e/0x20
 [151850.465343]  [<ffffffff811a0dea>] sys_newlstat+0x1a/0x40
 [151850.465350]  [<ffffffff816fdf5d>] system_call_fastpath+0x1a/0x1f
    
por Tombart 11.02.2014 / 12:03

1 resposta

3

A mensagem aparece no caso de um determinado processo (neste caso sftp-server ) não obter CPU por 120s (limite padrão). Isso pode ser causado por alta carga no sistema. Geralmente isso poderia ser causado em espera em qualquer recurso, provavelmente os candidatos são CPU, disco e rede.

Ao depurar esses problemas, você pode testar a velocidade de gravação no disco:

$ dd if=/dev/zero of=/tmp/output conv=fdatasync bs=384k count=1k; rm -f /tmp/output
1024+0 records in
1024+0 records out
402653184 bytes (403 MB) copied, 2753.13 s, 146 kB/s

Você deve esperar valores em MB / s, por ex. 50-230 MB/s dependendo do disco rígido. Nesse caso, o throughput 146 kB/s é extremamente baixo e pode causar essas mensagens do kernel.

Você também pode usar iostat -x 5 para monitorar a utilização do disco.

    
por 22.01.2016 / 20:23