# Nodes don't see each other on Google Container Engine **URL:** \ **Category:** Elasticsearch\ **Created:** [August 18, 2017, 12:50pm UTC](https://discuss.elastic.co/t/nodes-dont-see-each-other-on-google-container-engine/97582 "2017-08-18T12:50:34Z")\ **Posts on this page:** 2\ **Page:** 1
**Author:** ![Pablo\_Poveda](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/pablo_poveda/32/21262_2.png) [@Pablo\_Poveda](https://discuss.elastic.co/u/Pablo_Poveda)\ **Post date:** [August 18, 2017, 12:50pm UTC](https://discuss.elastic.co/t/nodes-dont-see-each-other-on-google-container-engine/97582/1 "2017-08-18T12:50:34Z")
Hello, I'm trying to run a 3 node cluster on Google Container Engine (GKE). I'm using Docker to run the images directly, not Kubernetes. The image is based on the one provided by [Google Cloud Launcher](https://github.com/GoogleCloudPlatform/elasticsearch-docker). Each node works correctly, but they are unable to see each other to form a cluster. The command I run is `docker run -p 9200-9400:9200-9400 image_name`. The firewall in Google is configured to allow traffic between ports 9100-9400. Current config file is ```auto node.name: ${HOSTNAME} network.host: _gce:hostname_ discovery.zen.minimum_master_nodes: 2 discovery.zen.ping.unicast.hosts: ["10.240.0.10:9300", "10.240.0.11:9300", "10.240.0.12:9300"] network.bind_host: localhost http.host: 0.0.0.0 readonlyrest: enable: false ``` I also tried using the discovery with the GCE Plugin, setting `transport.host` to the host IP instead of localhost (but that would result in the following error): ```auto org.elasticsearch.bootstrap.StartupException: BindTransportException[Failed to bind to [9300-9400]]; nested: BindEx ception[Cannot assign requested address]; ``` I tried all combinations of GCE provided hostnames, external and internal IPs for discovery and not setting the transport and host parameters, but that would result in a crash. I guess the problem lies somewhere with the `transport.host` parameter, but I've exhausted all options and copied all possible configuration I could find online. Output of running with the current `elasticsearch.yml`file is ```auto [2017-08-18T12:37:20,099][INFO][o.e.n.Node] [daa04be6c3b3] initializing ... [2017-08-18T12:37:20,314][INFO][o.e.e.NodeEnvironment] [daa04be6c3b3] using [1] data paths, mounts [[/usr/shar e/elasticsearch/data (/dev/sda1)]], net usable_space [23.2gb], net total_space [27.3gb], spins? [possibly], types [ ext4] [2017-08-18T12:37:20,316][INFO][o.e.e.NodeEnvironment] [daa04be6c3b3] heap size [6.3gb], compressed ordinary o bject pointers [true] [2017-08-18T12:37:20,318][INFO][o.e.n.Node] [daa04be6c3b3] node name [daa04be6c3b3], node ID [Zrndi Tk0SEyi6TSrTwI_pA] [2017-08-18T12:37:20,319][INFO][o.e.n.Node] [daa04be6c3b3] version[5.4.3], pid[1], build[eed30a8/20 17-06-22T00:34:03.743Z], OS[Linux/4.4.52+/amd64], JVM[Oracle Corporation/OpenJDK 64-Bit Server VM/1.8.0_131/25.131- b11] [2017-08-18T12:37:20,320][INFO][o.e.n.Node] [daa04be6c3b3] JVM arguments [-Xms6500m, -Xmx6500m, -XX :+UseConcMarkSweepGC, -XX:CMSInitiatingOccupancyFraction=75, -XX:+UseCMSInitiatingOccupancyOnly, -XX:+DisableExplic itGC, -XX:+AlwaysPreTouch, -Xss1m, -Djava.awt.headless=true, -Dfile.encoding=UTF-8, -Djna.nosys=true, -Djdk.io.perm issionsUseCanonicalPath=true, -Dio.netty.noUnsafe=true, -Dio.netty.noKeySetOptimization=true, -Dio.netty.recycler.m axCapacityPerThread=0, -Dlog4j.shutdownHookEnabled=false, -Dlog4j2.disable.jmx=true, -Dlog4j.skipJansi=true, -XX:+H eapDumpOnOutOfMemoryError, -Des.path.home=/usr/share/elasticsearch] [2017-08-18T12:37:22,484][INFO][o.e.p.PluginsService] [daa04be6c3b3] loaded module [aggs-matrix-stats] [2017-08-18T12:37:22,484][INFO][o.e.p.PluginsService] [daa04be6c3b3] loaded module [ingest-common] [2017-08-18T12:37:22,485][INFO][o.e.p.PluginsService] [daa04be6c3b3] loaded module [lang-expression] [2017-08-18T12:37:22,485][INFO][o.e.p.PluginsService] [daa04be6c3b3] loaded module [lang-groovy] [2017-08-18T12:37:22,485][INFO][o.e.p.PluginsService] [daa04be6c3b3] loaded module [lang-mustache] [2017-08-18T12:37:22,485][INFO][o.e.p.PluginsService] [daa04be6c3b3] loaded module [lang-painless] [2017-08-18T12:37:22,485][INFO][o.e.p.PluginsService] [daa04be6c3b3] loaded module [percolator] [2017-08-18T12:37:22,485][INFO][o.e.p.PluginsService] [daa04be6c3b3] loaded module [reindex] [2017-08-18T12:37:22,486][INFO][o.e.p.PluginsService] [daa04be6c3b3] loaded module [transport-netty3] [2017-08-18T12:37:22,486][INFO][o.e.p.PluginsService] [daa04be6c3b3] loaded module [transport-netty4] [2017-08-18T12:37:22,487][INFO][o.e.p.PluginsService] [daa04be6c3b3] loaded plugin [discovery-gce] [2017-08-18T12:37:22,487][INFO][o.e.p.PluginsService] [daa04be6c3b3] loaded plugin [readonlyrest] [2017-08-18T12:37:25,979][INFO][o.e.d.DiscoveryModule] [daa04be6c3b3] using discovery type [zen] SLF4J: Failed to load class "org.slf4j.impl.StaticLoggerBinder". SLF4J: Defaulting to no-operation (NOP) logger implementation SLF4J: See http://www.slf4j.org/codes.html#StaticLoggerBinder for further details. [2017-08-18T12:37:27,051][INFO][o.e.p.r.e.IndexLevelActionFilter] Configuration reloaded - ReadonlyREST disabled [2017-08-18T12:37:27,071][INFO][o.e.p.r.e.IndexLevelActionFilter] Readonly REST plugin was loaded... [2017-08-18T12:37:27,449][INFO][o.e.n.Node] [daa04be6c3b3] initialized [2017-08-18T12:37:27,450][INFO][o.e.n.Node] [daa04be6c3b3] starting ... [2017-08-18T12:37:27,937][INFO][o.e.t.TransportService] [daa04be6c3b3] publish_address {10.240.0.12:9300}, bound_addresses {127.0.0.1:9300}, {[::1]:9300} [2017-08-18T12:37:27,948][INFO][o.e.b.BootstrapChecks] [daa04be6c3b3] bound or publishing to a non-loopback or non-link-local address, enforcing bootstrap checks [2017-08-18T12:37:31,007][WARN][o.e.d.z.ZenDiscovery] [daa04be6c3b3] not enough master nodes discovered during pinging (found [[Candidate{node={daa04be6c3b3}{ZrndiTk0SEyi6TSrTwI_pA}{WXcRmBmxSIOSkGl_xLLNcQ}{gke-es5-cluster-de fault-pool-9896038c-bp3v.c.project.internal}{10.240.0.12:9300}, clusterStateVersion=-1}]], but needed [2]), pinging again ``` The `not enough master nodes` message then loops infinitely. A `curl` from one machine to another returns the standard `You Know, for Search` message for port 9200 and `url: (56) Recv failure: Connection reset by peer`for port 9300. Any help will be greatly appreciated. ---
**Author:** ![system](https://us1.discourse-cdn.com/elastic/original/3X/1/a/1ac57faf039f6b580b3f104ef42a2a89e41014de.png) [@system](https://discuss.elastic.co/u/system)\ **Post date:** [September 15, 2017, 12:50pm UTC](https://discuss.elastic.co/t/nodes-dont-see-each-other-on-google-container-engine/97582/2 "2017-09-15T12:50:59Z")
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.