Actions
Bug #36361
openupmap balancer: crash in calc_pg_upmaps if there's a destroyed osd in the tree
Status:
Need More Info
Priority:
Normal
Assignee:
-
Category:
balancer module
Target version:
-
% Done:
0%
Source:
Community (user)
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):
Description
If there is a "destroyed" osd in the crush tree, calc_pg_upmaps will crash:
2018-10-09 15:47:17.800180 7f0637355700 4 mgr[balancer] pools ['cinder-critical', 'volumes', 'test', 'images'] 2018-10-09 15:47:18.050334 7f0637355700 -1 *** Caught signal (Segmentation fault) ** in thread 7f0637355700 thread_name:balancer ceph version 12.2.8 (ae699615bac534ea496ee965ac6192cb7e0e07c0) luminous (stable) 1: (()+0x3f40c1) [0x5562647db0c1] 2: (()+0xf6d0) [0x7f064fb116d0] 3: (std::__detail::_List_node_base::_M_hook(std::__detail::_List_node_base*)+0xf) [0x7f064f43ee5f] 4: (CrushWrapper::_get_take_weight_osd_map(int, std::map<int, float, std::less<int>, std::allocator<std::pair<int const, float> > >*) const+0x20f) [0x556264ab0e6f] 5: (CrushWrapper::get_rule_weight_osd_map(unsigned int, std::map<int, float, std::less<int>, std::allocator<std::pair<int const, float> > >*) const+0x180) [0x556264ab1150] 6: (OSDMap::calc_pg_upmaps(CephContext*, float, int, std::set<long, std::less<long>, std::allocator<long> > const&, OSDMap::Incremental*)+0x4da) [0x55626490b18a] 7: (()+0x2e7974) [0x5562646ce974] 8: (PyEval_EvalFrameEx()+0x6df0) [0x7f0651a77cf0] 9: (PyEval_EvalCodeEx()+0x7ed) [0x7f0651a7a03d] 10: (PyEval_EvalFrameEx()+0x663c) [0x7f0651a7753c] 11: (PyEval_EvalFrameEx()+0x67bd) [0x7f0651a776bd] 12: (PyEval_EvalFrameEx()+0x67bd) [0x7f0651a776bd] 13: (PyEval_EvalCodeEx()+0x7ed) [0x7f0651a7a03d] 14: (()+0x70978) [0x7f0651a03978] 15: (PyObject_Call()+0x43) [0x7f06519dea63]
Here is the destroyed osd:
1455 hdd 5.45999 osd.1455 destroyed 0 1.00000
Updated by Dan van der Ster over 5 years ago
May in fact be related to or caused by http://tracker.ceph.com/issues/36378
Updated by Sebastian Wagner almost 5 years ago
- Category changed from ceph-mgr to balancer module
Updated by Konstantin Shalygin almost 3 years ago
- Backport deleted (
luminous,mimic)
Updated by Neha Ojha almost 3 years ago
- Status changed from New to Need More Info
Dan, have you seen this in newer versions?
Actions