Start Cloudera Manager Service to install JDK and Agent of the remaining three nodes
Continuing from the above, we start the remaining steps, start Cloudera Manager Service, and use this tool to automatically install jdk, and cloudera-agent on the remaining three nodes, server2,server3,server4
Before starting the cloudera-scm-server service, we can manually put the CDH package next week into the server1 host / opt/cloudera/parcel-repo directory to speed up the following steps.
Https://archive.cloudera.com/cdh7/6.0.1/parcels/
Look at the manifest.json file, find the hash value corresponding to your system, and make sure that the value in this CDH-6.0.1-1.cdh7.0.1.p0.590678-el7.parcel.sha file is consistent with that in manifest.json.
1. Start the cloudera-scm-server service
# systemctl start cloudera-scm-server
The waiting time is long. You can determine whether the startup is successful by querying the log.
2, login, http://192.168.1.201:7180/
User name: admin
Password: admin
The following individual pictures are transferred from the Internet. When editing this article, I stay in the installation and download agent interface, which is relatively slow.
3. The hash values mentioned above are the same, and the corresponding version will appear below.
4. The CDH version shown in the above figure is inconsistent, so it is in the process of downloading. After modifying the hash value, restart clouder-scm-server and start all over again. The correct version appears, which instantly prompts the download to be completed, and the installation process of 3 nodes appears.
5. Check the possible problems of the host and the solutions
Cloudera recommends setting / proc/sys/vm/swappiness to 10. Currently set to 60.
# echo 'vm.swappiness=10' > > / etc/sysctl.conf# sysctl-p
Large memory pages disabled
Echo never > / sys/kernel/mm/transparent_hugepage/defrag
Echo never > / sys/kernel/mm/transparent_hugepage/enabled
Or write
Vi / etc/rc.local
Upgrade software depends on version # yum install python-pip
# pip install-upgrade psycopg2
The host is in poor health.
Delete the cm_guid file under the agent directory and restart the agent service recovery of the failed node.
# cd / var/lib/cloudera-scm-agent/
# rm-rf cm_guid
# service cloudera-scm-agent restart
