After successful inspection we will know node UUID already, so no need
to do a proper lookup. In the future we might get rid of lookup completely
in this case.
Change-Id: Ic00049bf69f9ecc59ac4e389a3e6afe20ba67aba
This does a few things:
* Update hacking to the version in global-requirements. Old hacking was
installing a version of pbr that was breaking other packages.
* Fix all the hacking/pep8 rules that updating hacking raised.
* Do some general docstring cleanup, while already in there cleaning up
a bunch of docstrings due to H405 violations.
Change-Id: I1fc1e59d4c3d7b14631f8b576e3f3854bc452188
Closes-Bug: #1461717
Requests defaults to keeping 10 connections to the API alive. This
seems excessive, reduce it to 2.
Change-Id: Iac67fd4d855388d2671da3b7258b31d1eead77a9
This reverts commit 7b8b06c71401b5ce13fc73ed13b0e49e648cba8b.
The wrong function was fixed - it was the heartbeat, not the lookup,
that was checking for the wrong status code.
Change-Id: I35f49ca4fa1295d266f68b72b6ecdd5184181bd5
Allow configuration via command line arguments or kernel parameters.
Default to agent_ipmitool, the reference driver.
Depends on https://review.openstack.org/#/c/84795/12
Change-Id: I55c4a8713308d038002a6567471cd862bf89ec76
Bumping the payload version to two, and adding CPU, Memory, and Block
Device information to the inventory.
Added dependency on psutil for available memory and cpu_count.
Requirements line copied from global requirements
Change-Id: Ia39c85c91b1d60468667787a7978020084dc6c2a
This payload may change over time and should be versioned.
Corresponds to https://review.openstack.org/#/c/85228/
Change-Id: I58b1d69a2c41ff105038178632e289cb4b12c971
Using DynamicLoopingCall involved a few hacks to make it work properly. This
new BackOffLoopingCall will start an exponential backoff (with a configurable
jitter) when there is a failure. The backoff will continue until the given
function returns True or timeout is about to be exceeded. The function will
run indefinitely until either an exception is raised or timeout is reached.
I plan to merge this into oslo loopingcall and switch the heartbeat to this.
Change-Id: I1482348e98c6b68c34b3003645029e08135b1341
Currently, if a single lookup call to the Ironic API fails, the entire
agent errors out and restarts. This allows the agent to retry for a set
amount of time before throwing an uncaught exception forcing a restart.
Change-Id: I39752fb3f42ad3e4f15a49194f1554e1d3463cf8
Closes-Bug: 1297019
