- Stop using our custom
terraform-provider-ubuntu
plugin, since Terraform can now provide this functionality natively. - Expose
node_exporter
http on director. - Bump up to Terraform 0.8.3
- More version information from provisioning script and cli utils
- Adapt to use
kubeadm
1.6alpha line - Bump up to version 1.5.1 of Kubernetes
- Increase the number of worker instances from 2 to 4.
- Fix a bug where the border Squid proxy wouldn't stay running after zone spinup.
- Install and configure Prometheus
node_exporter
on each of our instances. - Add
quay.io
to our list of whitelisted container sources.
- Fixed a bug where
journal-remote
logs collected on the border node were not being cleaned automatically. This led to the border node running out of disk over long periods of time. This is a bug in systemd/journal-remote
, which we've worked around by cleaning up manually with a timer unit (aka cronjob).
- Disabled log forwarding from
journald
(which is good about not using up the entire disk) tosyslog
(which is not). This avoids an issue where logs could fill the root volume and prevent Kubernetes from functioning properly on a node. - Lowered the logging verbosity of
kubelet
to reduce pressure on our logging system.
- Our first v1.0.0 release candidate! We've made a ton of progress since our v0.0.x versions -- too much to reasonably list here. Suffice to say we built most of Substrate in the last year.
- Added a real CLI and a reworked build process.
- Begin work on new development flow, reintroduce
master
branch
- Remove proxy/squid setup
- Move kubernetes master to a single instance etcd setup
- Create a kubernetes worker ASG and launch configuration
- Remove Route53 setup for now
- Upgrade to Terraform 0.6.14.
- Upgrade to Kubernetes 1.2
- Made a test change to a naming thing to see if it gets picked up.
- Initial version of substrate with versioning/CI set up.