ceph同步数据过程OSD进程异常退出记录

h0945466

　　做的操作: ceph集群扩容了几个节点。

　　异常现象: ceph集群同步时，总是OSD进程异常的宕掉(同步一段时间数据后)。
　　ceph版本:  9.2.1
　　

　　日志:
7月 25 09:25:57 ceph6 ceph-osd[26051]: 0> 2017-07-25 09:25:57.471502 7f46fe478700 -1 common/HeartbeatMap.cc: In function 'bool ceph::HeartbeatMap::_ch
7月 25 09:25:57 ceph6 ceph-osd[26051]: common/HeartbeatMap.cc: 81: FAILED assert(0 == "hit suicide timeout")
7月 25 09:25:57 ceph6 ceph-osd[26051]: ceph version 9.2.1 (752b6a3020c3de74e07d2a8b4c5e48dab5a6b6fd)
7月 25 09:25:57 ceph6 ceph-osd[26051]: 1: (ceph::__ceph_assert_fail(char const*, char const*, int, char const*)+0x85) [0x7f47038330b5]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 2: (ceph::HeartbeatMap::_check(ceph::heartbeat_handle_d const*, char const*, long)+0x2d9) [0x7f47037728e9]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 3: (ceph::HeartbeatMap::is_healthy()+0xd6) [0x7f4703773126]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 4: (ceph::HeartbeatMap::check_touch_file()+0x2c) [0x7f47037738ec]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 5: (CephContextServiceThread::entry()+0x15b) [0x7f470384f2bb]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 6: (()+0x7dc5) [0x7f47018a6dc5]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 7: (clone()+0x6d) [0x7f470014f76d]
7月 25 09:25:57 ceph6 ceph-osd[26051]: NOTE: a copy of the executable, or `objdump -rdS ` is needed to interpret this.
7月 25 09:25:57 ceph6 ceph-osd[26051]: terminate called after throwing an instance of 'ceph::FailedAssertion'
7月 25 09:25:57 ceph6 ceph-osd[26051]: *** Caught signal (Aborted) **
7月 25 09:25:57 ceph6 ceph-osd[26051]: in thread 7f46fe478700
7月 25 09:25:57 ceph6 ceph-osd[26051]: ceph version 9.2.1 (752b6a3020c3de74e07d2a8b4c5e48dab5a6b6fd)
7月 25 09:25:57 ceph6 ceph-osd[26051]: 1: (()+0x7e6fe2) [0x7f470373dfe2]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 2: (()+0xf370) [0x7f47018ae370]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 3: (gsignal()+0x37) [0x7f470008d1d7]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 4: (abort()+0x148) [0x7f470008e8c8]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 5: (__gnu_cxx::__verbose_terminate_handler()+0x165) [0x7f47009919d5]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 6: (()+0x5e946) [0x7f470098f946]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 7: (()+0x5e973) [0x7f470098f973]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 8: (()+0x5eb9f) [0x7f470098fb9f]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 9: (ceph::__ceph_assert_fail(char const*, char const*, int, char const*)+0x27a) [0x7f47038332aa]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 10: (ceph::HeartbeatMap::_check(ceph::heartbeat_handle_d const*, char const*, long)+0x2d9) [0x7f47037728e9]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 11: (ceph::HeartbeatMap::is_healthy()+0xd6) [0x7f4703773126]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 12: (ceph::HeartbeatMap::check_touch_file()+0x2c) [0x7f47037738ec]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 13: (CephContextServiceThread::entry()+0x15b) [0x7f470384f2bb]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 14: (()+0x7dc5) [0x7f47018a6dc5]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 15: (clone()+0x6d) [0x7f470014f76d]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 2017-07-25 09:25:57.525027 7f46fe478700 -1 *** Caught signal (Aborted) **
7月 25 09:25:57 ceph6 ceph-osd[26051]: in thread 7f46fe478700
7月 25 09:25:57 ceph6 ceph-osd[26051]: ceph version 9.2.1 (752b6a3020c3de74e07d2a8b4c5e48dab5a6b6fd)
7月 25 09:25:57 ceph6 ceph-osd[26051]: 1: (()+0x7e6fe2) [0x7f470373dfe2]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 2: (()+0xf370) [0x7f47018ae370]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 3: (gsignal()+0x37) [0x7f470008d1d7]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 4: (abort()+0x148) [0x7f470008e8c8]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 5: (__gnu_cxx::__verbose_terminate_handler()+0x165) [0x7f47009919d5]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 6: (()+0x5e946) [0x7f470098f946]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 7: (()+0x5e973) [0x7f470098f973]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 8: (()+0x5eb9f) [0x7f470098fb9f]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 9: (ceph::__ceph_assert_fail(char const*, char const*, int, char const*)+0x27a) [0x7f47038332aa]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 10: (ceph::HeartbeatMap::_check(ceph::heartbeat_handle_d const*, char const*, long)+0x2d9) [0x7f47037728e9]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 11: (ceph::HeartbeatMap::is_healthy()+0xd6) [0x7f4703773126]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 12: (ceph::HeartbeatMap::check_touch_file()+0x2c) [0x7f47037738ec]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 13: (CephContextServiceThread::entry()+0x15b) [0x7f470384f2bb]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 14: (()+0x7dc5) [0x7f47018a6dc5]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 15: (clone()+0x6d) [0x7f470014f76d]
7月 25 09:25:57 ceph6 ceph-osd[26051]: NOTE: a copy of the executable, or `objdump -rdS ` is needed to interpret this.
7月 25 09:25:57 ceph6 ceph-osd[26051]: 0> 2017-07-25 09:25:57.525027 7f46fe478700 -1 *** Caught signal (Aborted) **
7月 25 09:25:57 ceph6 ceph-osd[26051]: in thread 7f46fe478700
7月 25 09:25:57 ceph6 ceph-osd[26051]: ceph version 9.2.1 (752b6a3020c3de74e07d2a8b4c5e48dab5a6b6fd)
7月 25 09:25:57 ceph6 ceph-osd[26051]: 1: (()+0x7e6fe2) [0x7f470373dfe2]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 2: (()+0xf370) [0x7f47018ae370]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 3: (gsignal()+0x37) [0x7f470008d1d7]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 4: (abort()+0x148) [0x7f470008e8c8]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 5: (__gnu_cxx::__verbose_terminate_handler()+0x165) [0x7f47009919d5]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 6: (()+0x5e946) [0x7f470098f946]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 7: (()+0x5e973) [0x7f470098f973]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 8: (()+0x5eb9f) [0x7f470098fb9f]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 9: (ceph::__ceph_assert_fail(char const*, char const*, int, char const*)+0x27a) [0x7f47038332aa]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 10: (ceph::HeartbeatMap::_check(ceph::heartbeat_handle_d const*, char const*, long)+0x2d9) [0x7f47037728e9]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 11: (ceph::HeartbeatMap::is_healthy()+0xd6) [0x7f4703773126]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 12: (ceph::HeartbeatMap::check_touch_file()+0x2c) [0x7f47037738ec]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 13: (CephContextServiceThread::entry()+0x15b) [0x7f470384f2bb]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 14: (()+0x7dc5) [0x7f47018a6dc5]
7月 25 09:25:57 ceph6 ceph-osd[26051]: 15: (clone()+0x6d) [0x7f470014f76d]
7月 25 09:25:57 ceph6 ceph-osd[26051]: NOTE: a copy of the executable, or `objdump -rdS ` is needed to interpret this.　　

　　用关键词"FAILED assert(0 == "hit suicide timeout")"搜索时，很多都为“common/HeartbeatMap.cc: 79”类开头的异常。
　　基本没有“common/HeartbeatMap.cc: 81”开头的。
　　

　　期间找了各种尝试，巴拉巴拉好多。最后发现问题出现在新节点与老节点内核版本不一致的情况。更新后恢复正常。