2
votes

I dont know how to create h2o multi node cluster? I tried to use flatfile but It still dose not work.

Please check the log file:

[root@ptdl-167 home]# java -Xmx10G -ea -jar h2o.jar -name VIPCLUS -flatfile flatfile.txt -ip 10.61.74. 202 -port 54231 07-27 09:03:42.476 10.61.74.202:54231 13206 main INFO: ----- H2O started ----- 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Build git branch: rel-vapnik 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Build git hash: 47dceae5c504ed6a2fad5de29292509d0b8024cc 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Build git describe: jenkins-master-3905-6-g47dceae 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Build project version: 3.12.0.1 (latest version: unknown) 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Build age: 1 month and 20 days 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Built by: 'jenkins' 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Built on: '2017-06-06 23:26:13' 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Watchdog Build git branch: (unknown) 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Watchdog Build git hash: (unknown) 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Watchdog Build git describe: (unknown) 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Watchdog Build project version: (unknown) 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Watchdog Built by: (unknown) 07-27 09:03:42.490 10.61.74.202:54231 13206 main INFO: Watchdog Built on: (unknown) 07-27 09:03:42.491 10.61.74.202:54231 13206 main INFO: Processed H2O arguments: [-name, VIPCLUS, -flatfile, flatfile.txt, -ip, 10.61.74.202, -port, 54231] 07-27 09:03:42.491 10.61.74.202:54231 13206 main INFO: Java availableProcessors: 8 07-27 09:03:42.491 10.61.74.202:54231 13206 main INFO: Java heap totalMemory: 238.0 MB 07-27 09:03:42.491 10.61.74.202:54231 13206 main INFO: Java heap maxMemory: 8.89 GB 07-27 09:03:42.491 10.61.74.202:54231 13206 main INFO: Java version: Java 1.8.0_121 (from Oracle Corporation) 07-27 09:03:42.491 10.61.74.202:54231 13206 main INFO: JVM launch parameters: [-Xmx10G, -ea] 07-27 09:03:42.491 10.61.74.202:54231 13206 main INFO: OS version: Linux 3.10.0-514.el7.x86_64 (amd64) 07-27 09:03:42.491 10.61.74.202:54231 13206 main INFO: Machine physical memory: 15.44 GB 07-27 09:03:42.491 10.61.74.202:54231 13206 main INFO: X-h2o-cluster-id: 1501121020922 07-27 09:03:42.491 10.61.74.202:54231 13206 main INFO: User name: 'root' 07-27 09:03:42.491 10.61.74.202:54231 13206 main INFO: IPv6 stack selected: false 07-27 09:03:42.492 10.61.74.202:54231 13206 main INFO: Possible IP Address: enp0s31f6 (enp0s31f6), fe80:0:0:0:f71d:d7d1:c1e9:a164%enp0s31f6 07-27 09:03:42.492 10.61.74.202:54231 13206 main INFO: Possible IP Address: enp0s31f6 (enp0s31f6), 10.61.74.202 07-27 09:03:42.492 10.61.74.202:54231 13206 main INFO: Possible IP Address: lo (lo), 0:0:0:0:0:0:0:1%lo 07-27 09:03:42.492 10.61.74.202:54231 13206 main INFO: Possible IP Address: lo (lo), 127.0.0.1 07-27 09:03:42.492 10.61.74.202:54231 13206 main INFO: H2O node running in unencrypted mode. 07-27 09:03:42.493 10.61.74.202:54231 13206 main INFO: Internal communication uses port: 54232 07-27 09:03:42.493 10.61.74.202:54231 13206 main INFO: Listening for HTTP and REST traffic on http://10.61.74. 202 :54231/ 07-27 09:03:42.494 10.61.74. 202:54231 13206 main WARN: Flatfile configuration does not include self: /10.61.74. 202:54231 but contains [/10.60.74.201:54231, /10.60.74.203:54231, /10.60.74.202:54231] 07-27 09:03:42.494 10.61.74.202:54231 13206 main INFO: H2O cloud name: 'VIPCLUS' on /10.61.74. 202:54231, static configuration based on -flatfile flatfile.txt 07-27 09:03:42.494 10.61.74.202:54231 13206 main INFO: If you have trouble connecting, try SSH tunneling from your local machine (e.g., via port 55555): 07-27 09:03:42.494 10.61.74.202:54231 13206 main INFO: 1. Open a terminal and run 'ssh -L 55555:localhost:54231 [email protected]' 07-27 09:03:42.494 10.61.74.202:54231 13206 main INFO: 2. Point your browser to http:// localhost :55555 07-27 09:03:42.495 10.61.74.202:54231 13206 main INFO: Log dir: '/tmp/h2o-root/h2ologs' 07-27 09:03:42.495 10.61.74.202:54231 13206 main INFO: Cur dir: '/home' 07-27 09:03:42.504 10.61.74.202:54231 13206 main INFO: HDFS subsystem successfully initialized 07-27 09:03:42.506 10.61.74.202:54231 13206 main INFO: S3 subsystem successfully initialized 07-27 09:03:42.506 10.61.74.202:54231 13206 main INFO: Flow dir: '/root/h2oflows' 07-27 09:03:42.513 10.61.74.202:54231 13206 main INFO: Cloud of size 1 formed [ptdl-167.viettel.com.vn/10.61.74.202:54231] 07-27 09:03:42.519 10.61.74.202:54231 13206 main INFO: Registered parsers: [GUESS, ARFF, XLS, SVMLight, AVRO, PARQUET, CSV] 07-27 09:03:42.519 10.61.74.202:54231 13206 main INFO: Watchdog extension initialized 07-27 09:03:42.519 10.61.74.202:54231 13206 main INFO: Registered 1 core extensions in: 1ms 07-27 09:03:42.519 10.61.74.202:54231 13206 main INFO: Registered H2O core extensions: [Watchdog] 07-27 09:03:42.659 10.61.74.202:54231 13206 main INFO: Registered: 160 REST APIs in: 140ms 07-27 09:03:42.660 10.61.74.202:54231 13206 main INFO: Registered REST API extensions: [XGBoost, water.api.RegisterResourceRoots, Core V3, Core V4, Algos, AutoML] 07-27 09:03:42.730 10.61.74.202:54231 13206 main INFO: Registered: 230 schemas in 70ms 07-27 09:03:42.731 10.61.74.202:54231 13206 main INFO: H2O started in 1806ms 07-27 09:03:42.731 10.61.74.202:54231 13206 main INFO: 07-27 09:03:42.731 10.61.74.202:54231 13206 main INFO: Open H2O Flow in your web browser: http://10.61.74.202:54231 07-27 09:03:42.731 10.61.74.202:54231 13206 main INFO: 07-27 09:04:30.291 10.61.74.202:54231 13206 #02:54231 ERRR: Got IO error when sending batch UDP bytes: java.net.NoRouteToHostException: No route to host 07-27 09:04:30.298 10.61.74.202:54231 13206 #01:54231 ERRR: Got IO error when sending batch UDP bytes: java.net.NoRouteToHostException: No route to host 07-27 09:04:30.298 10.61.74.202:54231 13206 #03:54231 ERRR: Got IO error when sending batch UDP bytes: java.net.NoRouteToHostException: No route to host

1

1 Answers

2
votes

You need to have all host running and sharing same flatfile

For example

  1. You have 3hosts, and create flatfile.txt for them: 10.0.0.6 10.0.0.7 10.0.0.8

  2. You need to distribute flatfile.txt around cluster (it is not done for free)

  3. On each host 10.0.0.{6,7,8} you need to run: java -jar h2o.jar -ip <HOST_IP> -flatfile flatfile.txt

Generally, this is manual process, better way is to use Hadoop/Yarn or Spark (and Sparkling Water).