Installation Stalled at "Running Network Connectivity and Throughput Tests..."

Ruben_Silva

Hello,

I new to Vertica. I''m trying to setup a aws cluster using the AMI at https://aws.amazon.com/amis/hp-vertica-analytics-platform-6-1-2-0-1374507686 and following the documentation at https://my.vertica.com/docs/6.1.x/PDF/HP_Vertica_6.1.x_Vertica_AWS.pdf.

I manage to get all the three instances running. However the following problems occurred:

#1. The manual is missing the configuration of an route in the Route Table: Destination=0.0.0.0/0 and Target the internet gateway (something like igw-xxx)

#2. The Amazon provided DNS does not seam to work, I had to manualy add to /etc/hosts the DNS entries. I do not intend to left this config like this but was the only way the install script allows me to continue the installation
#3. The command specified in the documentation does not seam to work I had to change it from:

/opt/vertica/sbin/install_vertica -s 10.0.3.100,10.0.3.101,10.0.3.98,10.0.3.99 -i ~/tst-key-pair.pem -L CE

to:

/opt/vertica/sbin/install_vertica -C -s 10.0.3.100,10.0.3.101,10.0.3.98,10.0.3.99 -i ~/tst-key-pair.pem -L ~/license.key -r vertica-6.1.2-0.x86_64.RHEL5.rpm

After all this tiny changes the install script stops at:

Running Network Connectivity and Throughput Tests...
Waiting for 1 of 3 sites... 

I have missed something in the instance setup? What can I do to trace back the root of the problem?

RS

Ruben_Silva

Ok I found another two issues with the documentation:

#4. The port 48073 used for the Connectivity and Throughput Tests is not listed to be opened in the security group, I ended up opening all internal traffic so don't know if any specific additional ports are needed

#5. Since amazon specifically states that no broadcasting is allowed (http://aws.amazon.com/vpc/faqs/#R4) the install_vertica script need the -T option set:

  [-T] direct transmit only: disallow UDP broadcast traffic

Hope this helps someone... I waste a lot of time with this bad documentation.

[Deleted User]

Hello,

Did you have the opportunity to run the /opt/vertica/bin/vnetperf validation script during your testing?
If so, can you share the results?

Thanks.

Nunzy_Antonelli

Thank you for being in touch with us.  We are reviewing our AWS documentation based upon your comments.   Some initial responses follow.

• Amazon has changed the way users find AMI versions.  Our latest AMI does not appear in search results.  We are working to correct this issue. In the meantime, you can find our latest AMIs with the links listed below. Please note that clicking on a link brings you to step 2 of the instance launch procedure within the Amazon interface (the AMI chosen is HP Vertica’s AMI for release 7.0.0-0).  The different links represent different regions where this AMIs can be launched.
  

https://console.aws.amazon.com/ec2/home?region=us-east-1#launchAmi=ami-21654848  (US East region - ami-21654848) 

https://console.aws.amazon.com/ec2/home?region=us-west-2#launchAmi=ami-5aeb8e6a  (US West region - ami-5aeb8e6a)

https://console.aws.amazon.com/ec2/home?region=ap-northeast-1#launchAmi=ami-3f3f5c3e  (Asia Pacific region - ami-3f3f5c3e)

• Because our latest AMI was not visible to you, you were using an earlier AMI with documentation intended for our later AMI.  Due to Amazon changes, and AMI enhancements, the documentation changes from one release to another.  You can find the latest documentation for our latest AMI here:

https://my.vertica.com/docs/7.0.x/HTML/index.htm#Authoring/UsingVerticaOnAWS/UsingHPVerticaOnAmazonWebServices.htm%3FTocPath%3DHP%C2%A0Vertica%20on%20Amazon%20Web%20Services%20Guide%7C_____0

We hope this response helps, and we regret the difficulties you encountered.

Ruben_Silva

Thank you for your comprehensive answer. Meanwhile I had found that AMI after receiving an email from Vertica announcing the new version. I can state that building a three node cluster was pretty straightforward. The update also solved a problem related with the machine name solving.

Ruben_Silva

Hello David, I did not run that script. As soon as I'm able to launch the cluster again (AWS has been complaining for insufficient capacity for a while now) I will share the results.

Ruben_Silva

Hello David, I did not run that script. As soon as I'm able to launch the cluster again (AWS has been complaining for insufficient capacity for a while now) I will share the results.