-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/52581/
-----------------------------------------------------------
Review request for Aurora, John Sirois and Zameer Manji.
Repository: aurora
Description
-------
This was more a random experiment on my end. We use this value in production as
our clusters are rather small and it can happen that `#instances in a job >>
#agents`, so a smaller filter duration increases scheduling performance. Turns
out this is also slightly helpful in our e2e tests:
Before the patch:
real 27m52.457s
user 0m2.308s
sys 0m0.938s
with the patch:
real 23m56.899s
user 0m2.277s
sys 0m0.884s
Both experiments were preceeded with a `vagrant destroy && vagrant up`.
The improvement is minimal and could probably also due to other external
factors as I have made only one run each.
Diffs
-----
examples/vagrant/upstart/aurora-scheduler-kerberos.conf
f1852f1daac6c5379de34af713580ae4c4615548
examples/vagrant/upstart/aurora-scheduler.conf
91b27d7a7b3d9bfb9ccec29b52f5e0adb2e078f5
Diff: https://reviews.apache.org/r/52581/diff/
Testing
-------
Thanks,
Stephan Erb
