67be896e0f
This patch is based upon a downstream patch which came up in discussion amongst the ironic community when some operators began discussing a case where resource providers had disappeared from a running deployment with several thousand baremetal nodes. Discussion amongst operators and developers ensued and we were able to determine that this was still an issue in the current upstream code and that time difference between collecting data and then reconciling the records was a source of the issue. Per Arun, they have been running this change downstream and had not seen any reoccurances of the issue since the patch was applied. This patch was originally authored by Arun S A G, and below is his original commit mesage. An instance could be launched and scheduled to a compute node between get_uuids_by_host() call and _get_node_list() call. If that happens the ironic node.instance_uuid may not be None but the instance_uuid will be missing from the instance list returned by get_uuids_by_host() method. This is possible because _get_node_list() takes several minutes to return in large baremetal clusters and a lot can happen in that time. This causes the compute node to be orphaned and associated resource provider to be deleted from placement. Once the resource provider is deleted it is never created again until the service restarts. Since resource provider is deleted subsequent boots/rebuilds to the same host will fail. This behaviour is visibile in VMbooter nodes because it constantly launches and deletes instances there by increasing the likelihood of this race condition happening in large ironic clusters. To reduce the chance of this race condition we call _get_node_list() first followed by get_uuids_by_host() method. Change-Id: I55bde8dd33154e17bbdb3c4b0e7a83a20e8487e8 Co-Authored-By: Arun S A G <saga@yahoo-inc.com> Related-Bug: #1841481 (cherry picked from commit |
||
|---|---|---|
| .. | ||
| notes | ||
| source | ||