v1.5.0 / 2024-03-04

What's Changed

• Update go version to 1.20 by @cristinaleonr in #44
• Remove state for retired sites by @nkinkade in #46

New Contributors

• @cristinaleonr made their first contribution in #44

Full Changelog: v1.4.1...v1.5.0

v1.4.1

What's Changed

• Change master to main in .travis.yml by @SaiedKazemi in #42
• Migrate builds from Travis-CI to Cloud Build by @nkinkade in #43

New Contributors

• @SaiedKazemi made their first contribution in #42

Full Changelog: v1.4.0...v1.4.1

v1.4.0

What's Changed

• Adds a new siteinfo package by @nkinkade in #40
• Integrates new "sites" package into the rest of GMX by @nkinkade in #41

Full Changelog: v1.3.3...v1.4.0

Adds `site` label to all machine metrics

v1.3.2...v1.3.3

Removes v1 node name metrics

As a facility for easing the transition from v1 to v2 node names, GMX exporter metrics for both v1 and v2 names. Now that we are 100% v2 names on the platform it is no longer necessary to publish metrics for v1 names.

Bug fix release

Fixes issue #36.

Makes GMX compatible with v2 node names

Merge pull request #35 from m-lab/sandbox-kinkade

Makes GMX compatible with v2 node names.

Weekly release

Fixes a bug in which Prometheus metrics for production nodes were leaking into the staging instance of GMX, and vice versa.

Completely refactors code + fixes a few bugs

The primary feature of this release is a complete refactor of the code by @pboothe. A lot of the code is now broken out into packages, all of the code uses better Golang best practices, testing coverage is complete, etc.

Two bugs were also fixed in this release.

New --project flag, "node" label for all machine metrics + bugfix

• Adds a new --project flag to GMX which is used to filter webhooks by project in the standard manner. Since we only have a single m-lab/ops-tracker repository, the webhook has no way of knowing whether to send a webhook to a given GMX instance in each project, meaning that each GMX instance in each project will maintain maintenance metrics for all machines, no matter which project they are in, which ends up giving us spurious entries in mlab-ns queries for the project. This new feature should resolve that.

• Adds a new node label to every machine maintenance metric. Sometimes we need to join the gmx_machine_maintenance metrics on the machine label, and other times on the node label. This change will obviate the need for replace_labels in Prometheus to add a node label that is identical to the machine label. They will still be identical, but GMX will now do this for you.

• Fixes a bug that we didn't previously even know existed in which maintenance could be entered for the same machine/site and Github issue multiple times, meaning that when we went to exit maintenance for that issue, only one instance of the maintenance would be removed, effectively leaving the machine/site in maintenance.