[openstack-dev] [all] Proposal to change integrated neutron grenade gate job to multi-node

Brian Haley haleyb.dev at gmail.com
Fri Jul 14 21:05:57 UTC 2017


Hi,

While looking at ways to reduce the number of jobs we run in the Neutron 
gate, I found we ran two very similar jobs for some projects:

gate-grenade-dsvm-neutron-ubuntu-xenial (single-node job)
gate-grenade-dsvm-neutron-multinode-ubuntu-xenial (2-node job)

We talked about this in the Neutron CI meeting this week [1] and felt it 
best to remove the single-node job and just run the multi-node job, 
mostly because it more mimics a "real" Neutron deployment where there 
are separate controller and compute nodes.  Looking at the Neutron 
grafana dashboard [2] the two jobs have about the same failure rate in 
the gate (~0), so I don't think there will be any problems with the switch.

This has an impact on the integrated gate since it currently runs the 
single-node job, so I wanted ot get thoughts on any issues they'd have 
with this change [3].

Thanks,

-Brian

[1] 
http://eavesdrop.openstack.org/meetings/neutron_ci/2017/neutron_ci.2017-07-11-16.00.log.html#l-112
[3] http://grafana.openstack.org/dashboard/db/neutron-failure-rate
[2] https://review.openstack.org/#/c/483600/



More information about the OpenStack-dev mailing list