Getting API throttled (?) by Elastic, cannot understand why

Elastic Search
1

Hi.

We have an Elastic Cloud 7.17.18 managed instance we use for our staging app, along with two others we use for Training and Production.

In the staging environment we’re rather frequently unable to connect to our Elasticsearch instance due to connection refused or similar errors. It really looks like a throttling mechanism, but we’re unable to identify the cause of this blocking, as we’re not doing anything that could lead to this behavior.

Well, of course this could be accidental, and we don’t know (yet). What I’m looking for, here, is help on identifying proofs of this blocks, if it can happen in Elastic and where should I look for such errors.

I tried in Monitoring > Logs and metrics, Access and security > Network security, I see no error logs or other evidence.

Where should I find if our app is being blocked for too many attempts, or for unkind usage?

As a background, our app is PHP8.2 based, using Elastica library to connect over HTTPS.

The error happens at:

Elastica\Exception\Connection\HttpException Couldn't connect to host, Elasticsearch down? 
    vendor/ruflin/elastica/src/Transport/Http.php:186 Elastica\Transport\Http::exec
    vendor/ruflin/elastica/src/Request.php:183 Elastica\Request::send
    vendor/ruflin/elastica/src/Client.php:545 Elastica\Client::request
    vendor/ruflin/elastica/src/Client.php:580 Elastica\Client::requestEndpoint
    vendor/ruflin/elastica/src/Cluster.php:60 Elastica\Cluster::refresh
    vendor/ruflin/elastica/src/Cluster.php:48 Elastica\Cluster::__construct

thanks

2

Hi @maxxer

In general Elastic Cloud does not throttle API requests... There are some throttles on the Elastic Coud Hosted Admin APIs itself such as billing requests and things like that.

So you'll have to provide more information about exactly which API you are hitting

And kind of what the volume you're hitting.

If you are hitting normal elasticsearch endpoints APIs and they're not responding, that's probably more a function of the capacity of the cluster.

We have clusters that are ingesting a quarter million events per second or 10s of thousands search requests on a properly sized cluster

So provide a little more information and maybe we can help... Be precise on exactly what API endpoint you're hitting and what you're trying to accomplish and what the volume is

All that said... There certainly could be an issue with something along the chain

Also, keep in mind since it's elastic cloud, depending on your subscription level, you'd be able to open a support ticket.

3

(post deleted by author)

4

welcome to the forum @Danny_M

But did you intend to post this on this thread?

5

No I'm playing around with Claude and must've posted it on accident sorry…. Nice that tho

1 Like