[Lustre-discuss] Client evictions

[Nikolay Kvetsinski]

Hello,
we have a lot of client evictions. Here are the errors on the OSS:

LustreError: 138-a: lustre0-OST0001: A client on nid 192.168.12.38 at tcp was
evicted due to a lock glimpse callback time out: rc -107

On the MDS :

LustreError: 138-a: lustre0-MDT0000: A client on nid 192.168.12.38 at tcp was
evicted due to a lock blocking callback time out: rc -107

On the client :

INFO: task ldlm_bl_01:2818 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
ldlm_bl_01    D 0000000000000002     0  2818      2 0x00000080
 ffff88080460fd50 0000000000000046 0000000000000000 ffffffffa0e5a967
 0000000100000000 0000000000000000 ffff88080460fcf0 ffffffffa0e59f22
 ffff88080460dab8 ffff88080460ffd8 000000000000fb88 ffff88080460dab8
Call Trace:
 [<ffffffffa0e5a967>] ? cfs_hash_bd_lookup_intent+0x37/0x130 [libcfs]
 [<ffffffffa0e59f22>] ? cfs_hash_bd_add_locked+0x62/0x90 [libcfs]
 [<ffffffff8150f1ee>] __mutex_lock_slowpath+0x13e/0x180
 [<ffffffff8150f08b>] mutex_lock+0x2b/0x50
 [<ffffffffa0fc6ecf>] cl_lock_mutex_get+0x6f/0xd0 [obdclass]
 [<ffffffffa133695a>] osc_ldlm_blocking_ast+0x7a/0x350 [osc]
 [<ffffffffa10e4f00>] ldlm_handle_bl_callback+0x130/0x400 [ptlrpc]
 [<ffffffffa10e5451>] ldlm_bl_thread_main+0x281/0x3d0 [ptlrpc]
 [<ffffffff81063310>] ? default_wake_function+0x0/0x20
 [<ffffffffa10e51d0>] ? ldlm_bl_thread_main+0x0/0x3d0 [ptlrpc]
 [<ffffffff8100c0ca>] child_rip+0xa/0x20
 [<ffffffffa10e51d0>] ? ldlm_bl_thread_main+0x0/0x3d0 [ptlrpc]
 [<ffffffffa10e51d0>] ? ldlm_bl_thread_main+0x0/0x3d0 [ptlrpc]
 [<ffffffff8100c0c0>] ? child_rip+0x0/0x20
INFO: task nautilus:3741 blocked for more than 120 seconds.


Can you help me with somethig... just for start. Lustre 2.4 , OS - CentOS
6.4

Cheers,
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20130911/9d06b6c1/attachment.htm>