release-1.2

This release makes a couple of very minor changes:

• The HTTP-server sets up a default timeout of 300seconds / 5 minutes.
• The display of logs on the report-view shows where the log game from, in the manifest.
  • Thanks to @obi11235 for this contribution.

release-1.1

This release improves some of the internal details to handle error-conditions, and avoid them. There are no real user-visible changes, instead this release is about internal cleanup and tooling support.

It is now possible to submit metrics to a remote carbon server via the use of the golang-metrics library.

We've also moved from using go-bindata to my own tool for embedding resources into the binary, implant.

release-1.0.3

This release, unfortunately, drops binaries for different OS and architectures. This is necessary because the previous binaries just didn't work - each would complain about the lack of CGO being present for the SQLite3 library.

In terms of functional changes we now only regard nodes as orphaned if they've not been seen for three and a half days, rather than 24 hours. It is also possible to prune orphaned hosts, via "puppet-summary prune -orphaned"

release-1.0.1

This release updates the handling of the node view:

• Each report now has an ID listed, which corresponds to the point on the runtime-graph.
• Clicking upon a point in the graph-view will take you to the appropriate entry.

These changes are designed to allow a user to quickly identify a particularly slow/fast report, as documented in #45.

(Release 1.0.1 is a brief update to release-1.0 (withdrawn) which fixes a test-case failure I had not observed. Sorry for the churn!)

release-0.9

This release includes a couple of small fixups:

• The project contains a Dockerfile for simpler installation under docker.
• Reports which contain no changes can be pruned, via
  • puppet-summary prune -unchanged [-days 7]
  • Records of runs are still kept, but the report files are removed from disc and the UI doesn't contain links to the missing report.
• To ease running in a container a new flag -auto-prune has been added to the serve command.
  • Running "puppet-summary serve -auto-prune [flags]" will prune/reap reports older than 7 days, once per week, without the need to add a cronjob.

release-0.8

This release improves the display of the report-view, such that file-edits and similar are pre-formatted, which should make them easier to read.

This change was submitted by @matejzero - thanks :)

release-0.7

This release updates the handling of the end-point /api/state/${state}, allowing the caller to choose whether they prefer JSON, plain-text, or XML as output.

Allowing the caller to choose their preferred output-type matches our other end-points nicely, and has been documented in our API-guide.

release 0.6

This release continues to make improvements to the test-cases included in the source-repository, currently the coverage is hovering around 70%.

Beyond fixing some minor issues, and improving test-coverage, this new release introduces a new facility:

• Nodes are now searchable.

Every page of the UI now includes a search-box, allowing you to quickly & easily show nodes matching a given pattern.

release-0.5

This release improves the code quality, both in terms of the report card and the internal comments of the code-base.

The various views which are presented are now also fully scriptable, each view will return either:

• HTML
  • For viewing in your browser
• application/json
  • Which can be used for scripting-purposes
• application/xml
  • Which can be used for scripting-purposes

This support for returning different content-types, is fully documented in our API documentation. Our hacking guide has also been updated to demonstrate how you might cross-compile this code, useful if you wish to deploy upon a Raspberry Pi, for example.

release-0.4

This release introduces the display of "orphaned" nodes, nodes which have previously been heard from but which haven't submitted a report within a day.

It also introduces the "radiator-view", which is discoverable at the foot of ever page. Perfect for large status-screens.