Have indices in Elasticsearch? This is the tool for you!
Like a museum curator manages the exhibits and collections on display, Elasticsearch Curator helps you curate, or manage your indices.
| Version | ES 1.x | AWS ES 1.x | ES 2.x | AWS ES 2.x | ES 5.x | AWS ES 5.x |
|---|---|---|---|---|---|---|
| 3 | yes | yes* | yes | yes* | no | no |
| 4 | no | no | yes | no | yes | no |
| 5 | no | no | no | no | yes | yes* |
It is important to note that Curator 4 will not work with indices created in versions of Elasticsearch older than 1.4 (if they have been subsequently re-indexed, they will work). This is because those older indices lack index metadata that Curator 4 requires. Curator 4 will simply exclude any such indices from being acted on, and you will get a warning message like the following:
2016-07-31 10:36:17,423 WARNING Index: YOUR_INDEX_NAME has no "creation_date"! This implies that the index predates Elasticsearch v1.4. For safety, this index will be removed from the actionable list.
It is also important to note that Curator 4 requires access to the
/_cluster/state/metadata endpoint. Forks of Elasticsearch which do not
support this endpoint (such as AWS ES, see #717) will not be able to use
Curator version 4.
* It appears that AWS ES does not allow access to the snapshot status endpoint for the 1.x, 2.x, 5.1, and 5.3 versions. This prevents Curator 3 from being used to make snapshots.
? Curator 4 and 5 should work with AWS ES 5.x, but the
/_cluster/state/metadata endpoint is still not fully supported (see #880).
If a future patch fixes this, then Curator 4 and 5 should work with AWS ES 5.x.
| Branch | Status |
|---|---|
| Master |  |
| 5.x |  |
| 5.1 |  |
| 5.0 |  |
| 4.x |  |
| 4.3 |  |
PyPI: 
Version 5 of Curator ships with both an API and a wrapper script (which is actually defined as an entry point). The API allows you to write your own scripts to accomplish similar goals, or even new and different things with the Curator API, and the Elasticsearch Python API.
The Curator CLI Documentation is now a part of the document repository at http://elastic.co/guide at http://www.elastic.co/guide/en/elasticsearch/client/curator/current/index.html
See the Installation guide and the command-line usage guide
Running curator --help will also show usage information.
Encountering issues like DistributionNotFound? See the FAQ for that issue, and more.
The documentation for the CLI is now part of the document repository at http://elastic.co/guide at http://www.elastic.co/guide/en/elasticsearch/client/curator/current/index.html
The Curator Wiki on Github is now a place to add your own examples and ideas.
- fork the repo
- make changes in your fork
- add tests to cover your changes (if necessary)
- run tests
- sign the CLA
- send a pull request!
To run from source, use the run_curator.py script in the root directory of
the project.
To run the test suite just run python setup.py test
When changing code, contributing new code or fixing a bug please make sure you include tests in your PR (or mark it as without tests so that someone else can pick it up to add the tests). When fixing a bug please make sure the test actually tests the bug - it should fail without the code changes and pass after they're applied (it can still be one commit of course).
The tests will try to connect to your local elasticsearch instance and run
integration tests against it. This will delete all the data stored there! You
can use the env variable TEST_ES_SERVER to point to a different instance
(for example, 'otherhost:9203').
The combination of setuptools and cx_Freeze allows for Curator to be compiled into binary packages. These consist of a binary file placed in a directory which contains all the libraries required to run it.
In order to make a binary package you must manually install the cx_freeze
python module. You can do this via pip, or python setup.py install,
or by package, if such exists for your platform. In order to make it compile on
recent Debian/Ubuntu platforms, a patch had to be applied to the setup.py
file in the extracted folder. This patch file is in the unix_packages
directory in this repository.
With cx_freeze installed, building a binary package is as simple as running
python setup.py build_exe. In Linux distributions, the results will be in
the build directory, in a subdirectory labelled
exe.linux-x86_64-${PYVER}, where ${PYVER} is the current major/minor
version of Python, e.g. 2.7. This directory can be renamed as desired.
Other entry-points that are defined in the setup.py file, such as
es_repo_mgr, will also appear in this directory.
The process is identical for building the binary package for Windows. It must
be run from a Windows machine with all dependencies installed. Executables in
Windows will have the .exe suffix attached. The directory in build will
be named exe.win-amd64-${PYVER}, where ${PYVER} is the current major/minor
version of Python, e.g. 2.7. This directory can be renamed as desired.
In Windows, cx_Freeze also allows for building rudimentary MSI installers. This
can be done by invoking python setup.py bdist_msi. The MSI fill will be in
the dist directory, and will be named
elasticsearch-curator-#.#.#-amd64.msi, where the major, minor, and patch
version numbers are substituted accordingly. One drawback to this rudimentary
MSI is that it does not allow updates to be installed on top of the existing
installation. You must uninstall the old version before installing the newer
one.
The unix_packages directory contains the build_packages.sh script used
to generate the packages for the Curator YUM and APT repositories. The
Vagrant directory has the Vagrantfiles used in conjunction with the
build_packages.sh script. If you wish to use this method on your own, you
must ensure that the shared folders exist. /curator_packages is where the
packages will be placed after building. /curator_source is the path to the
Curator source code, so that the build_packages.sh script can be called from
there. The build_packages.sh script does not use the local source code,
but rather pulls the version specified as an argument directly from GitHub.
Version 5 of Curator is the current master branch. It supports only 5.x
versions of Elasticsearch.
Curator was first called clearESindices.py [1] and was almost immediately
renamed to logstash_index_cleaner.py [1]. After a time it was migrated under
the [logstash](https://github.com/elastic/logstash) repository as
expire_logs. Soon thereafter, Jordan Sissel was hired by Elasticsearch, as
was the original author of this tool. It became Elasticsearch Curator after
that and is now hosted at <https://github.com/elastic/curator>