Hi,
I want to know how can we incorporate curator in ILM so that I can take a backup of an index when it's rolled over.
You can’t incorporate Curator in ILM. The new SLM (Snapshot Lifecycle Management) works independently of ILM and does not filter snapshots like you are describing either, but will snapshot based on index names and wildcards.
You can, however, run Curator alongside ILM. As presently constituted, using SLM alongside Curator for snapshots could cause collisions that prevent snapshots from happening, so I would recommend just using SLM and consider that solved. If, however, you are determined to snapshot how and when you want, and to filter to snapshot only rolled-over indices, then Curator can do that by using multiple filters: one to select the entire index pattern, and one to eliminate the is_write_index index (the current “live” index). This will snapshot all rolled indices and skip the active one.
Do remember that segments already in the repository (from indices which have already had at least one snapshot) will not be re-copied. That means that snapshotting all indices but the active one will only copy data that is not yet in the repository.
Hi Aaron,
Out of interest, is there a better way to do this than to just exclude the latest index of that pattern?
Thanks.
If there is, I’m not sure what it is. The example can be further extended. For example, after selecting an entire index pattern and eliminating the currently active one, you could further limit the remaining indices by date or by count.
What is it you would like to see?
I was just wondering if there was a way to query the is_write_index status of an index or something, but querying based on count should be absolutely fine. Thanks for all the work you do on Curator by the way 
There's already an issue for that feature. I will probably get to it sooner, rather than later, as there appear to be more interested parties.
It's trickier to implement than you would think, though. In order to maintain reverse-compatibility with older versions of Elasticsearch I have to detect what version of ES is running, and exit/abort with an error code if someone tries to use it on an incompatible release of Elasticsearch.
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.