Merge lp:~ken-pepple/nova/lp718360 into lp:~hudson-openstack/nova/trunk
Status: | Rejected |
---|---|
Rejected by: | Vish Ishaya |
Proposed branch: | lp:~ken-pepple/nova/lp718360 |
Merge into: | lp:~hudson-openstack/nova/trunk |
Diff against target: |
40 lines (+7/-2) 2 files modified
nova/crypto.py (+4/-1) nova/utils.py (+3/-1) |
To merge this branch: | bzr merge lp:~ken-pepple/nova/lp718360 |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Todd Willey (community) | Disapprove | ||
Review via email: mp+49919@code.launchpad.net |
Commit message
fixed debug level logging in nova/utils.py
Description of the change
this is tricky and a bit brittle -- the problem we are seeing is that when you run nova-compute, you are importing nova/utils.py before you've parsed FLAGS, so you can't set the correct LOG.level if you've added the --verbose flag (it always stays at the default 20 / INFO value). To fix, I've brought in flags to utils.py and had it parse the sys.argv ... which will work most of the time. The one case were it will not work is when you've let haven't specified flags or a flagfile on the command line.
Also updated crypto.py and rpc.py, which exhibited similar problems.
Unmerged revisions
- 675. By Ken Pepple
-
removed unnecessary flag parse
- 674. By Ken Pepple
-
initialized FLAGS with sys.argv for LOG.debug to work
- 673. By Ken Pepple
-
fixed (?) logging issues in rpc and utils
- 672. By Ken Pepple
-
merged trunk
Is there anywhere else FLAGS() is being called in library code instead of the binaries? I'm thinking it may make testing more brittle, and should probably be avoided.