Weekly release

• Upgrades CoreOS and k8s components to more recent versions.
• Reconfigures iPXE retry logic to use a shorter delay between retries, and to reboot the machine after max retries has been reached (instead of falling back to a shell).

Weekly release

• Add drac.ipv4 to the stage1 kernel arguments.
• Add periodic job and timer for cron-like jobs.
  • Currently the only periodic job is one looking for hung/stuck host containerd-shim processes.
• Add authorized_keys for root user to stage1 images for a fallback login in case the DRAC is inaccessible and the stage1 boot fails for some reason.
• Configure basic DRAC settings during stage1 boot (user, passwd, IPv4 configs).

Build images for site HKG01

Removes the 2nd IPv6 DNS nameserver entry. systemd is complaining abo…

…ut too many configured nameserver (more than 3) and kubelet is constantly complaining too ('Nameserver limits were exceeded') (#144)

Builds images for DEL02 and MAA02

Removes the 2nd IPv6 DNS nameserver entry. systemd is complaining abo…

…ut too many configured nameserver (more than 3) and kubelet is constantly complaining too ('Nameserver limits were exceeded') (#144)

Removes 2nd IPv6 DNS nameserver + builds several images

Adding a 2nd IPv6 DNS nameserver (for a total of 4 (2 IPv4 + 2 IPv6)) was causing systemd-resolved to complain about too many configured nameservers, and kubectl too. This release removes the 2nd IPv6 nameserver.

This release should also build updated images for GRU04 and TRN02, both of which have the same IPv4 subnet, but updated IPv6 prefixes.

Builds images for new site VIE02

VIE01 is becoming VIE02 due to a change in the IPv6 prefix for VIE01 from TOP-IX.

Introduces separate builds for stage1 and stage3 artifacts

This release introduces separate builds for stage1 and stage3 artifacts. Now, building stage3 artifacts (which is fast) is done in its own build, and building stage1 artifacts (very slow) happens separately. The reason for this is that often changes we make to the epoxy-images repository only change stage3 artifacts, and having to wait literally hours for identical stage1 artifacts to build blocks progress.

Weekly release

This release features:

• Updates CoreOS version to latest stable (2135.6.0) (reboot required).
• Configures TCP pacing on boot based on site's uplink speed (reboot required).

Builds images for new site GRU01 (Sao Paulo)

Configure platform cluster nodes with IPv6 addresses (#133)

* Updates generate_network_config.sh to also configure IPv6 on the node.

* Changes awk field separator from : to ,.

* Removes redunant prefix length from the end of IPv4 and IPv6 addresses.

* Fixes a small, bad reference in a comment.

Increases CloudBuild disk to 400GB

The build in mlab-oti failed for want of disk space even at 200GB, so this release ups the CB VM's disk space to 400GB.