So long as we are referencing the version of Kibana in our readme, we
cannot reliably automate the bumping of the Kibana version as part of
our release process.
None of the other repos in the stack include version information or
links to snapshots in the readme. If folks want to know what version
they're looking at, the package.json is the source of truth.
Kibana should be able to run against Elasticsearch nodes that are at the
same minor version regardless of whether they are at the same patch
version. In the event that their patch versions differ, we render a
warning.
- Remove engineVersion from elasticsearch plugin config.
- Use Kibana package.json version instead.
- Use new rules, documented in README.
- Log warning if ES is newer than Kibana.
- Add isEsCompatibleWithKibana and kibanaVersion.
- Remove versionSatisfies.
We do not release zip archives for any operating system except windows,
and we do not release tar.gz archives for windows. For consistency and
clarity sake, we'll explicitly list the architecture (x86) of the
windows build as well.
These changes are necessary for Kibana to be compatible with Elastic's
unified release process from 5.0 onward. The way artifacts get created
has not changed, but the naming conventions have.
The default behavior of the build task is to now apply the -snapshot
suffix dynamically rather than us manually hardcoding and managing it
within the source code itself. The `--release` flag will drop the
-snapshot suffix on a build, which should be used for any release
candidate.
The default behavior of the build task has also changed to create
rpm/deb packages as well. Since we've only confirmed that this works on
linux, you can override that behavior by passing `skip-os-packages`.
If you do not want to create any zip or tar.gz archives, you can pass
`--skip-archives`.
tar.gz is sufficient for all operating systems except for windows, which by
default only includes zip. There's no reason to provide both links, and we
intend to stop publishing the alternative links altogether at some point.
Since the "use event times to create index names" functionality for
index patterns is now deprecated in favor of the automatic optimizations
powered by the field stats api, this tip is no longer necessary.