Created on 01-20-2016 06:01 PM - edited 09-16-2022 02:58 AM
Hi there,
I'm trying to work with AWS for HDPCD practice exam. I ran " hdfs dfs -ls / " cmd on the terminal and getting " connection refused error". I could see all the services are not running and displays with red colour rectangle icon. I ran the below command
ssh root@namenode
password: hadoop
service ambari-agent restart
exit
~/start-all-services.sh
but, no services are running except Zookeeper and Ganglia services. Please let me know to solve this issue. Thanks! @rich
Created 01-20-2016 08:54 PM
If you startup one of our practice exams and the cluster services are not running, here are some things to try first:
1. Run the start_all_services.sh script. You can watch the progress of this script in Ambari - it uses the Ambari REST API to start all HDP services.
2. Login to Ambari and try to start services that way
3. Restart the ambari-agent process on each node in the cluster (password for root is hadoop), then run start_all_services.sh again:
ssh root@namenode "ambari-agent restart" ssh root@resourcemanager "ambari-agent restart" # This only applies the HDPCA exam ssh root@hiveserver "ambari-agent restart" # This only applies to the HDPCA exam
If none of the steps above fix the issue, then it is likely an issue with the Docker containers not starting properly. Since you do not have root access on your practice exam, the quickest solution is to simply terminate your EC2 instance and startup another one.
-Rich
Created 01-20-2016 06:02 PM
@rich Tagging
Created 01-20-2016 06:02 PM
@Jeeva Jeeva Thanks for tagging @rich
Created 01-20-2016 06:03 PM
you need to manage services via Ambari UI http://hostname:8080 with user "admin" password "admin". Start services that way not through script. @Jeeva Jeeva
Created 01-20-2016 06:17 PM
Hi Artem Ervits ,
Thanks for your reply.. I tried through ambari starting the services, but doesn't work...
Created 01-20-2016 06:26 PM
Email me your public dns and I can take a look. certification@hortonworks.com
Created 01-20-2016 06:49 PM
@rich Please publish an article on the fix for future references.
Created 01-20-2016 07:16 PM
Hi Rich,
Thanks for your reply, the public dns changes every time whenever i login to. If you are available now, i'll login to aws and send you the dns. Please let me know if you are available now. @rich
Created 01-20-2016 08:54 PM
If you startup one of our practice exams and the cluster services are not running, here are some things to try first:
1. Run the start_all_services.sh script. You can watch the progress of this script in Ambari - it uses the Ambari REST API to start all HDP services.
2. Login to Ambari and try to start services that way
3. Restart the ambari-agent process on each node in the cluster (password for root is hadoop), then run start_all_services.sh again:
ssh root@namenode "ambari-agent restart" ssh root@resourcemanager "ambari-agent restart" # This only applies the HDPCA exam ssh root@hiveserver "ambari-agent restart" # This only applies to the HDPCA exam
If none of the steps above fix the issue, then it is likely an issue with the Docker containers not starting properly. Since you do not have root access on your practice exam, the quickest solution is to simply terminate your EC2 instance and startup another one.
-Rich
Created 01-21-2016 05:29 PM
Hi there,
@rich
@Guilherme Braccialli
I'm have scheduled my HDPCD exam next week. I couldn't practice exam in AWS since there are no services running except Ganglia and ZooKeeper. I tried the below commands, but no use.
ssh root@namenode ##password is: hadoop service ambari-agent restartexit./start_all_services.sh I even tried through ambari to start the services, i can see the progress in green and ultimately ends with red color triangle. Please help me, so that i can practice the exam prior to my original one. I checked even if firewall status is turned off.If you're available, please try to login the instance and see the error. Thanks in advance... @rich
Created 01-21-2016 05:45 PM
Like I said in the answer above, if restarting services fails then the quickest and easiest fix is to terminate the EC2 instance and start a new one. Have you done that yet? I can't login to your instance - I do not know the IP address.
Created 01-21-2016 05:58 PM
Hi thanks for your reply. I terminated the old one and this is the new one and still the same problem.
Public DNS: ec2-54-86-20-155.compute-1.amazonaws.com
Public IP 54.86.20.155 Could you please login and see what the issue is ? Thanks..Created 01-21-2016 06:18 PM
Hi rich, I greatly appreciate your help. Thanks for your support. I'd like to know what the issue is for future reference.. Once again Thanks..
Created 01-22-2016 01:33 AM
The correct answer is up above - about turning off safemode on the namenode
Created 01-21-2016 06:20 PM
@rich Is it possible to write an article on this topic ? @Mark Herring
Created 01-21-2016 07:00 PM
I explained the fix above and the steps involved, but my answer got out of order in this discussion. It's the answer about shutting off safemode on the NN.
Created 01-21-2016 06:13 PM
Hi @Jeeva Jeeva : it looks like your NameNode was in safemode. Leaving safemode allowed your cluster to startup properly. Here is what I did - then ran start_all_services.sh and everything started up fine:
$ ssh root@namenode (password is "hadoop") # su - hdfs $ hdfs dfsadmin -safemode leave
Created 04-16-2016 06:55 AM
ec2-54-186-154-189.us-west-2.compute.amazonaws.com.. am also facing this issue.. please help me
Created 04-16-2016 11:18 AM
1. Restart ambari-agent
2. Run the start_all_services.sh script
3. Wait 5 minutes and see if the cluster is up and running
If those steps fail, then terminate your EC2 instance and build a new one. That will be quicker than trying to debug the issue.
Created 04-16-2016 03:03 PM
I am also facing the same issue, cam you please help. I tried all the options in the blog. Twice I rebuild the cluster as well. I am not able to start & work.
Created 04-16-2016 03:26 PM
Which Amazon region are getting the AMI from? Perhaps there is an AMI that is not working. Try starting up one in a different region and see if that works.