Ben Nemec 2f80b51e53 Cache server status ... In a large environment with a lot of bmcs running, just the periodic status checks from something like Ironic can generate a lot of api traffic on the host cloud. Since nothing should be changing the status of the instances out from under us, we can cache the status and only look it up when we know it may have changed. This should reduce the number of api calls significantly. For reference, it looks like Ironic checks power status once per minute by default. In an environment with 5 baremetal nodes per baremetal stack and 50 total stacks (not a particularly huge environment), this can generate 250 server gets per minute, which seems to have the knock-on effect of generating a lot of Neutron traffic too. My assumption is that Nova is retrieving network details for the server each time the server is retrieved, and in a net-iso stack this involves a lot of networks and ports.		2016-08-09 12:23:32 -05:00
..
tests	Cache server status	2016-08-09 12:23:32 -05:00
__init__.py	Reorganize into package and add tox for testing	2016-07-14 12:23:41 -05:00
build_nodes_json.py	Add message about node files	2016-07-18 15:52:41 -05:00
deploy.py	Reorganize into package and add tox for testing	2016-07-14 12:23:41 -05:00
openstackbmc.py	Cache server status	2016-08-09 12:23:32 -05:00