Hi, we have a custom Elasticsearch deployment using marathon. The 3 data nodes are deployed using Docker containerizer on the DC/OS 9.x.x.x network and have a VIP of es6data.marathon.l4lb.thisdcos.directory:9200.
We are also deploying Logstash using Docker containerizer on bridge network. When Logstash tries to connect to es6data.marathon.l4lb.thisdcos.directory:9200 it fails 1 out of 3 times. This can be tested by executing curl inside the running Logstash container using docker exec -ti xxxxxx bash. It can be tested with curl es6data.marathon.l4lb.thisdcos.directory:9200 or curl 9.x.x.x.x and the connection of the local node will fail.
This only fails inside the docker container running in bridge mode. On the actual physical host slave agent, there's no connectivity issues.