Any way to configure keepAlive to reuse connections ?

[juanviamonte]

We love this project! is there any ways to: Any way to configure keepAlive to reuse connections ?

Thansk a lot

[andrykonchin]

I am not aware of any explicit configuration option to manage the Keep-Alive header in the AWS SDK.

But looks like AWS SDK re-uses connection by default. There is a pool of HTTP clients and the http_idle_timeout option that declares how much time not used HTTP client is kept in the pool (documentation).

Moreover, looks like this option is used also to set the Keep-Alive header (for the Net::HTTP client):

https://github.com/aws/aws-sdk-ruby/blob/9557e95a0b72a3c324e81443237745d9b6e3f3e8/gems/aws-sdk-core/lib/seahorse/client/net_http/connection_pool.rb#L284

I haven't checked manually whether TCP connections are re-used by sequenced API calls to DynamoDB but looks like they should, out of the box.

[andrykonchin]

I have checked manually and found out that connections are reused in scope of a single Ruby process within configured http_idle_timeout interval. So TCP-connection to some service (and region) can be reused by any other client object.

[juanviamonte]

@andrykonchin thanks a lot