Add checkpoint logging when building an instance in compute manager
When the large ops job fails because instances timeout during the build,
it's hard to know what is going on for a given instance in the compute
manager's _do_build_and_run_instance flow since a lot of the trace
logging that happens is in the virt drivers.
The large ops job uses the fake virt driver which doesn't do much
logging at all. Rather than add logging to the fake virt driver, it'd be
useful to add some logging to the compute manager when we are doing
sub-tasks like building networks, building bdms, and spawning the instance
on the hypervisor.
The 'Starting instance' log message is changed to debug level and the
completed task logging unit of work messages with time spent are done at
info level.
Reviewed: https:/ /review. openstack. org/228980 /git.openstack. org/cgit/ openstack/ nova/commit/ ?id=9bb4c07b299 6aa9e3153e6d6f2 8e7aa600c64100
Committed: https:/
Submitter: Jenkins
Branch: master
commit 9bb4c07b2996aa9 e3153e6d6f28e7a a600c64100
Author: Matt Riedemann <email address hidden>
Date: Tue Sep 29 07:21:19 2015 -0700
Add checkpoint logging when building an instance in compute manager
When the large ops job fails because instances timeout during the build, and_run_ instance flow since a lot of the trace
it's hard to know what is going on for a given instance in the compute
manager's _do_build_
logging that happens is in the virt drivers.
The large ops job uses the fake virt driver which doesn't do much
logging at all. Rather than add logging to the fake virt driver, it'd be
useful to add some logging to the compute manager when we are doing
sub-tasks like building networks, building bdms, and spawning the instance
on the hypervisor.
Per: https:/ /wiki.openstack .org/wiki/ LoggingStandard s
The 'Starting instance' log message is changed to debug level and the
completed task logging unit of work messages with time spent are done at
info level.
Related-Bug: #1500615
Change-Id: I6b9e7fc7b3a14c cf8b1531fa08ce8 d732e23775a