Activity log for bug #1606825

Date Who What changed Old value New value Message
2016-07-27 09:06:53 Roman Podoliaka bug added bug
2016-07-27 09:07:26 Roman Podoliaka tags mos-nova
2016-07-27 09:07:30 Roman Podoliaka tags mos-nova area-nova
2016-07-27 09:07:35 Roman Podoliaka mos: assignee Roman Podoliaka (rpodolyaka)
2016-07-27 09:07:39 Roman Podoliaka mos: milestone 10.0
2016-07-27 09:07:50 Roman Podoliaka mos: status New Confirmed
2016-07-27 09:07:59 Roman Podoliaka mos: importance Undecided High
2016-07-27 09:08:05 Roman Podoliaka nominated for series mos/9.x
2016-07-27 09:08:05 Roman Podoliaka bug task added mos/9.x
2016-07-27 09:08:12 Roman Podoliaka nominated for series mos/10.0.x
2016-07-27 09:08:12 Roman Podoliaka bug task added mos/10.0.x
2016-07-27 09:08:17 Roman Podoliaka mos/10.0.x: status New Confirmed
2016-07-27 09:08:19 Roman Podoliaka mos/10.0.x: importance Undecided High
2016-07-27 09:08:20 Roman Podoliaka mos/10.0.x: assignee Roman Podoliaka (rpodolyaka)
2016-07-27 09:08:23 Roman Podoliaka mos/10.0.x: milestone 10.0
2016-07-27 09:08:25 Roman Podoliaka mos/9.x: milestone 10.0 9.1
2016-07-27 10:21:17 Roman Podoliaka attachment added a process core dump, python2.7 ubuntu 16.04 https://bugs.launchpad.net/mos/+bug/1606825/+attachment/4708049/+files/core.dump.gz
2016-07-27 10:28:19 Roman Podoliaka description While executing a call to librbd nova-compute can hang for a while and eventually go down in nova service-list output. strace'ing shows that a process is stuck on acquiring a mutex: root@node-153:~# strace -p 16675 Process 16675 attached futex(0x7fff084ce36c, FUTEX_WAIT_PRIVATE, 1, NULL gdb allows to see the traceback: http://paste.openstack.org/show/542101/ ^ which basically means calls to librbd (C library) are not monkey-patched and do not allow switch context to another green thread in a eventlet-based process. To fix this we need something similar to: https://review.openstack.org/#/c/223077/ While executing a call to librbd nova-compute can hang for a while and eventually go down in nova service-list output. strace'ing shows that a process is stuck on acquiring a mutex: root@node-153:~# strace -p 16675 Process 16675 attached futex(0x7fff084ce36c, FUTEX_WAIT_PRIVATE, 1, NULL gdb allows to see the traceback: http://paste.openstack.org/show/542534/ ^ which basically means calls to librbd (C library) are not monkey-patched and do not allow switch context to another green thread in a eventlet-based process. To fix this we need something similar to: https://review.openstack.org/#/c/223077/
2016-07-28 17:06:29 Roman Podoliaka description While executing a call to librbd nova-compute can hang for a while and eventually go down in nova service-list output. strace'ing shows that a process is stuck on acquiring a mutex: root@node-153:~# strace -p 16675 Process 16675 attached futex(0x7fff084ce36c, FUTEX_WAIT_PRIVATE, 1, NULL gdb allows to see the traceback: http://paste.openstack.org/show/542534/ ^ which basically means calls to librbd (C library) are not monkey-patched and do not allow switch context to another green thread in a eventlet-based process. To fix this we need something similar to: https://review.openstack.org/#/c/223077/ While executing a call to librbd nova-compute may hang for a while and eventually go down in nova service-list output. strace'ing shows that a process is stuck on acquiring a mutex: root@node-153:~# strace -p 16675 Process 16675 attached futex(0x7fff084ce36c, FUTEX_WAIT_PRIVATE, 1, NULL gdb allows to see the traceback: http://paste.openstack.org/show/542534/ ^ which basically means calls to librbd (C library) are not monkey-patched and do not allow to switch the execution context to another green thread in a eventlet-based process. To avoid blocking of the whole nova-compute process on calls to librbd we should wrap them with tpool.execute() (http://eventlet.net/doc/threading.html#eventlet.tpool.execute)
2016-07-28 17:06:55 Roman Podoliaka description While executing a call to librbd nova-compute may hang for a while and eventually go down in nova service-list output. strace'ing shows that a process is stuck on acquiring a mutex: root@node-153:~# strace -p 16675 Process 16675 attached futex(0x7fff084ce36c, FUTEX_WAIT_PRIVATE, 1, NULL gdb allows to see the traceback: http://paste.openstack.org/show/542534/ ^ which basically means calls to librbd (C library) are not monkey-patched and do not allow to switch the execution context to another green thread in a eventlet-based process. To avoid blocking of the whole nova-compute process on calls to librbd we should wrap them with tpool.execute() (http://eventlet.net/doc/threading.html#eventlet.tpool.execute) Upstream bug: https://bugs.launchpad.net/nova/+bug/1607461 While executing a call to librbd nova-compute may hang for a while and eventually go down in nova service-list output. strace'ing shows that a process is stuck on acquiring a mutex: root@node-153:~# strace -p 16675 Process 16675 attached futex(0x7fff084ce36c, FUTEX_WAIT_PRIVATE, 1, NULL gdb allows to see the traceback: http://paste.openstack.org/show/542534/ ^ which basically means calls to librbd (C library) are not monkey-patched and do not allow to switch the execution context to another green thread in a eventlet-based process. To avoid blocking of the whole nova-compute process on calls to librbd we should wrap them with tpool.execute() (http://eventlet.net/doc/threading.html#eventlet.tpool.execute)
2016-09-09 10:33:16 Roman Podoliaka mos/9.x: status Confirmed Fix Committed
2016-09-09 10:33:52 Roman Podoliaka mos/10.0.x: status Confirmed Fix Committed
2016-09-13 11:16:01 Timur Nurlygayanov mos/10.0.x: status Fix Committed Fix Released
2016-09-13 11:16:04 Timur Nurlygayanov mos/9.x: status Fix Committed Fix Released