Yes. You'll be able to configure devoted learn nodes for the domains. When picking a dedicated learn configuration, you could specify the instance type and occasion count.
The full depend of HTTP requests created to OpenSearch Dashboards. If your procedure is gradual or you see large numbers of Dashboards requests, think about expanding the scale on the instance sort.
For each-node metric for the amount of occasions an exception happened when seeking to load a graph to the cache. This metric is just appropriate to approximate k-NN search.
The quantity of rejected tasks while in the search thread pool. If this range frequently grows, consider scaling your cluster.
The most percentage in the Java heap useful for all data nodes within the cluster. OpenSearch Services makes use of 50 % of an instance's RAM with the Java heap, up to a heap measurement of 32 GiB.
The overall amount of requests rejected on primary shards as a consequence of distant storage and replication stress. This is certainly calculated ranging from the final OpenSearch Services approach startup.
In this post, we talk about two alternatives to attain this: the Terraform OpenSearch supplier plus the Evolution library. Which 1 is best suited to your use case is dependent upon the tooling you happen to be accustomed to, your language of selection, and also your existing pipeline.
Cross-cluster lookup is just not supported within the t2 and m3 relatives scenarios resulting from specialized limitation.
If 5 of those shards are on a single node, the node would report 5 for this metric, Although the shopper only produced one ask for.
Automate the generation and administration of alerts & dashboards through API, such as the muting of alerts for the duration of servicing intervals
The most proportion of your Java heap employed for all devoted coordinator nodes inside the cluster. We advise going to a larger instance sort when this metric reaches eighty five %.
The amount of enter and output (I/O) OpenSearch monitoring operations for every second for examine operations on EBS volumes when micro-bursting is taken into consideration.
To support significant clusters with overall instance count of more than two hundred across data and heat nodes, We've got a 10K Restrict on the amount of indexes returned through the _cat/segments API.
The number of follower indexes which have been effectively produced by a replication rule for a selected link. AutoFollowNumFailedStartReplication