Skip to content

ODC Migration

Compare
Choose a tag to compare
@cscarp cscarp released this 21 Mar 16:55
· 416 commits to main since this release
dad2819

This release mainly addresses the new FABRIC ODC Rack where the Infrastructure central monitoring systems will be running.
ODC now hosts MinIO (replacing CEPH), central metrics and Thanos compactor.
Changes to Central Metrics role, along with the additional roles for minio, compactor and ping-plus are now implemented using docker-compose instead of Python Docker SDK. This results in more compact installation roles with more autonomy from the other roles. Common tasks are still used for creating directory structures and uploading configuration files. Common files are also still used for NGINX conf files and alerts.
Ping-plus is an addition to enable pings and other Blackbox Exporter probes to be made from the UKY datacenter rack which has access to the operations network. Formerly these probes were made from the Central metrics VM.
Other changes:
Grafana updated to version 9
Thanos updated to v0.29.0
Simple index page added to metrics.fabric-testbed.net
cAdvisor is being used on new ODC VMs and is expected to replace other instances of Docker Exporter as Docker Exporter is no longer being updated.
Nginx Proxy added for Thanos web UI's for status/debuging.
passlib no longer needs to be installed with bcrypt option.
Added promtheus_src for non-head node instances of Prometheus.
Thanos Stores updated to 23 racks.
GENI roles have been removed.
Dev/test install playbooks have been updated or removed.