I/O reactor status : STOPPED

[Zixell]

Version: 7.1.0

We are having these errors occasionally, for some reason ES client after fail killing IO Reactor and it is not recovering, so the only thing we can do to fix it - is restarting our service.

I've tried to implement custom exception handler for IO reactor and pass it to httpClientBuilder but in our case it brakes our SSL config and causes for problems then it solves.

override val restClient: RestClient = RestClient
    .builder(elasticHosts: _*)
    .setRequestConfigCallback(NoOpRequestConfigCallback)
    .setHttpClientConfigCallback(new HttpClientConfigCallback {
      override def customizeHttpClient(httpClientBuilder: HttpAsyncClientBuilder): HttpAsyncClientBuilder = {
        if (withBasicAuth) httpClientBuilder.setDefaultCredentialsProvider(getCredentialsProvider)
        if (isASecureConnection) httpClientBuilder.setSSLContext(getContextJDK(configuration))
        httpClientBuilder.setDefaultIOReactorConfig(IOReactorConfig.custom()
          .setSoKeepAlive(true)
          .build())
      }
    })
    .build()

  override val client: ElasticClient = ElasticClient(JavaClient.fromRestClient(restClient))

This is out rest client configuration.

Anyway, my question is: is it a known issue and if so, maybe it was fixed in newer versions? I found somewhere that it is was fixed in 7.16.3 but could find any evidence of that.

Feel free to ask any other information! As I found online many people experiencing the same issue, but I couldn't find solid solution for this, so I decided to create an Issue.

[Philippus]

I don't think this is an issue with elastic4s, but maybe with Elasticsearch itself?