Is there any way to sync Kibanas across multiple clusters? Essentially, a way to sync all visualizations, dashboards, queries, etc. from one Kibana to another. I'm aware that you can use remote reindex, and/or Kibana's export API, but all of those are only good for creating missing elements. What I mean by that is I can replicate non-existent dashboards, visualizations, queries and so on, but I can't update them the same way after creation.
What I would like to know is if it's possible to completely mirror one Kibana to another via some way of automation that does not require the second Kibana to have downtime. This would include replications of any changes to existing dashboards, removing some saved queries, and so on. If something like that isn't solveable with the Basic license, is it solved in a payed one?
Thanks in advance for any help.
Hi @lduvnjak ,
There is official documentation available regarding multiple instances configuration
My understanding is that it should solve your request, otherwise please provide more context on how you organise your kibana & ES clusters.
Thanks for the quick reply. This documentation is for multiple Kibana instances on one cluster.
What we have are two separate ES clusters, with the same data. So we have two Elastic clusters, two Kibanas, which are not, and cannot be clustered.
However, even though the Elastic clusters are separate, because they are supposed to be identical replicas, we need everything relating to Kibana to be mirrored as well.
Let me know if you have any more questions.
Thanks for clarifying. I don't think we have a solution for self managed to mirror clusters the way you described. As you mentioned, you can try and automated continuous saved objects export/import, but I'm not sure how well maintained this approach will be.
It is solved in Elastic Cloud, where we make sure to have a snapshot (mirror) and be responsible for cluster availability.
My teammate told me snapshot and restore can be done on local, but only if you can handle the next kibana instance being down for a bit, as the index needs closing before restoring a snapshot.
Thank you for the quick and concrete replies. I figured it wouldn't be possible without a bit of downtime, but just wanted to confirm.
Thanks a bunch, and have a wonderful day!
You can use cross cluster replication, but it might be a bit heavy for this.
Although this does seem possible for an active-passive arhitecture, is it also possible for active-active? Can one index be a leader on both clusters and be replicated bi-directionaly?
One more question - since this is a system indice, and the default superuser role no longer has access to them, do you need to create a new role with
allow_restricted_indices: true, and use that role for CCR?
Edit - CCR limitations mention that it only works for user created indices, not system indices:
Cross-cluster replication | Elasticsearch Guide [8.5] | Elastic
Ahh damn, sorry about that!
No worries, any suggestion is always greatly appreciated !
We'll leave it at - it's possible, although with a bit of downtime, and a custom solution involving either a remote reindex or snapshot/restore.