The logging in the scheduler is what takes most of the time,
once you are doing no DB calls.
This ensures the servicegroup is_up and retry_filter only log
on failures, which is more in line with the other filters.
Using the test "test_performance_check_select_destination"
modifying it to use 200 hosts, the request time went from
70ms to under 10ms.
We could reduce logging further, but this minimal change makes
a massive impact. The next steps would require a config value
to disable all debug logging in the scheduler.
Reviewed: https:/ /review. openstack. org/84806 /git.openstack. org/cgit/ openstack/ nova/commit/ ?id=93f29574c41 6aebcd1bd7527d3 007664aa073766
Committed: https:/
Submitter: Jenkins
Branch: master
commit 93f29574c416aeb cd1bd7527d30076 64aa073766
Author: John Garbutt <email address hidden>
Date: Wed Apr 2 17:37:03 2014 +0100
Reduce logging in scheduler
The logging in the scheduler is what takes most of the time,
once you are doing no DB calls.
This ensures the servicegroup is_up and retry_filter only log
on failures, which is more in line with the other filters.
Using the test "test_performan ce_check_ select_ destination"
modifying it to use 200 hosts, the request time went from
70ms to under 10ms.
We could reduce logging further, but this minimal change makes
a massive impact. The next steps would require a config value
to disable all debug logging in the scheduler.
Closes-Bug: #1301515
Change-Id: Ia4c7e506e8df9c 560c349e2ebec3f 5111aa0cb0d