Bug #22527
closedmgr.x runtime_error in json_spirit on upgrade
0%
Description
2017-12-21T00:20:45.577 INFO:tasks.ceph.mgr.x.smithi024.stderr:terminate called after throwing an instance of 'std::runtime_error' 2017-12-21T00:20:45.591 INFO:tasks.ceph.mgr.x.smithi024.stderr: what(): value type is 1 not 0 2017-12-21T00:20:45.591 INFO:tasks.ceph.mgr.x.smithi024.stderr:*** Caught signal (Aborted) ** 2017-12-21T00:20:45.591 INFO:tasks.ceph.mgr.x.smithi024.stderr: in thread 7feb0f469700 thread_name:fn_anonymous 2017-12-21T00:20:45.593 INFO:tasks.ceph.mgr.x.smithi024.stderr: ceph version 12.2.2-39-g88b4534 (88b4534a0fab6f7020874ff8903b2f2eb9d06eb2) luminous (stable) 2017-12-21T00:20:45.594 INFO:tasks.ceph.mgr.x.smithi024.stderr: 1: (()+0x3f40a4) [0x55635edb80a4] 2017-12-21T00:20:45.594 INFO:tasks.ceph.mgr.x.smithi024.stderr: 2: (()+0x11390) [0x7feb1792f390] 2017-12-21T00:20:45.594 INFO:tasks.ceph.mgr.x.smithi024.stderr: 3: (gsignal()+0x38) [0x7feb168bf428] 2017-12-21T00:20:45.594 INFO:tasks.ceph.mgr.x.smithi024.stderr: 4: (abort()+0x16a) [0x7feb168c102a] 2017-12-21T00:20:45.594 INFO:tasks.ceph.mgr.x.smithi024.stderr: 5: (__gnu_cxx::__verbose_terminate_handler()+0x16d) [0x7feb1720284d] 2017-12-21T00:20:45.594 INFO:tasks.ceph.mgr.x.smithi024.stderr: 6: (()+0x8d6b6) [0x7feb172006b6] 2017-12-21T00:20:45.595 INFO:tasks.ceph.mgr.x.smithi024.stderr: 7: (()+0x8d701) [0x7feb17200701] 2017-12-21T00:20:45.595 INFO:tasks.ceph.mgr.x.smithi024.stderr: 8: (()+0x8d919) [0x7feb17200919] 2017-12-21T00:20:45.595 INFO:tasks.ceph.mgr.x.smithi024.stderr: 9: (json_spirit::Value_impl<json_spirit::Config_map<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > >::check_type(json_spirit::Value_type) const+0x12d) [0x55635ec9611d] 2017-12-21T00:20:45.595 INFO:tasks.ceph.mgr.x.smithi024.stderr: 10: (json_spirit::Value_impl<json_spirit::Config_map<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > >::get_obj()+0xb) [0x55635ecc607b] 2017-12-21T00:20:45.595 INFO:tasks.ceph.mgr.x.smithi024.stderr: 11: (MetadataUpdate::finish(int)+0x502) [0x55635ecc7592] 2017-12-21T00:20:45.596 INFO:tasks.ceph.mgr.x.smithi024.stderr: 12: (Context::complete(int)+0x9) [0x55635ec6fc39] 2017-12-21T00:20:45.596 INFO:tasks.ceph.mgr.x.smithi024.stderr: 13: (Finisher::finisher_thread_entry()+0x460) [0x55635edf80d0] 2017-12-21T00:20:45.596 INFO:tasks.ceph.mgr.x.smithi024.stderr: 14: (()+0x76ba) [0x7feb179256ba] 2017-12-21T00:20:45.596 INFO:tasks.ceph.mgr.x.smithi024.stderr: 15: (clone()+0x6d) [0x7feb169913dd] 2017-12-21T00:20:45.596 INFO:tasks.ceph.mgr.x.smithi024.stderr:2017-12-21 00:20:45.588061 7feb0f469700 -1 *** Caught signal (Aborted) ** 2017-12-21T00:20:45.596 INFO:tasks.ceph.mgr.x.smithi024.stderr: in thread 7feb0f469700 thread_name:fn_anonymous
/a/sage-2017-12-20_22:44:45-rados-wip-sage-testing-2017-12-20-1357-distro-basic-smithi/1986414
Updated by Kefu Chai over 6 years ago
/a/kchai-2017-12-23_05:58:55-rados-wip-kefu-testing-2017-12-23-1241-distro-basic-mira/1996509
Updated by Kefu Chai over 6 years ago
it's caused by https://github.com/ceph/ceph/pull/19369. also this change breaks the upgrade test in a different way: http://pulpito.ceph.com/kchai-2017-12-24_02:20:05-rados-wip-kefu-testing-2017-12-23-1241-distro-basic-mira/1996875/, and https://github.com/ceph/ceph/pull/19369#commitcomment-26456668
Updated by Kefu Chai over 6 years ago
- Project changed from mgr to CephFS
- Component(FS) MDSMonitor added
Updated by Kefu Chai over 6 years ago
- Project changed from CephFS to mgr
- Assignee set to Patrick Donnelly
hey Patrick, could you take a look?
Updated by Patrick Donnelly over 6 years ago
I'm inclined to revert [1] as it's not worth the trouble it's caused. There isn't a clean way to make the upgrade testing work and changing the JSON format broke the mgr.
Updated by John Spray over 6 years ago
+1 -- I'd revert that. The original format with the dicts is kind of ugly, but I don't think there's a strong enough need to spend effort managing the transition to lists (at least, unless someone else cares enough to do the whole job including dealing with upgrades from the mgr side).
Updated by Patrick Donnelly over 6 years ago
- Status changed from 12 to Resolved
Fixed by f6e1a797d45d7d646835bad8e258908407b2695b.
Updated by Kefu Chai over 6 years ago
i still have this issue in a upgrade test. but the branch being tested does include the revert change. weird enough..
Updated by Patrick Donnelly over 6 years ago
- Related to Bug #21393: MDSMonitor: inconsistent role/who usage in command help added
Updated by Patrick Donnelly over 6 years ago
Kefu Chai wrote:
i still have this issue in a upgrade test. but the branch being tested does include the revert change. weird enough..
Caused by the fix to #21393
Updated by Patrick Donnelly over 6 years ago
- Status changed from Resolved to Fix Under Review
Patrick Donnelly wrote:
Kefu Chai wrote:
i still have this issue in a upgrade test. but the branch being tested does include the revert change. weird enough..
Caused by the fix to #21393
Updated by Sage Weil over 6 years ago
- Status changed from Fix Under Review to Resolved