Troubleshooting Seed Node Installation

Describes some common issues that can interfere with seed node installation.

Issue: Fabric Creation Fails If Host-Name Resolution Takes Too Long

Fabric creation requires host-name resolution to complete in less than 300 ms. If host-name resolution takes longer than 300 ms, fabric creation can fail. To check the speed of host-name resolution on a node, enter one of these commands:
  • time nslookup <hostname>
  • time getent ahosts <hostname>
Host-name resolution depends on the name servers that you are using. You can find them in the /etc/resolv.conf file. If you cannot use faster name servers, try improving the resolution time by providing local resolution. You can do this by updating the/etc/hosts file on each of the fabric nodes with the details of the other fabric nodes. For example:
<node1IP> <node1fqdn> <node1hostname>
<node2IP> <node2fqdn> <node2hostname>
<node3IP> <node3fqdn> <node3hostname>

Issue: Seed Node Container Services Do Not Come Up

Use these steps to resolve the problem:
  1. After running the script to deploy the container for the Data Fabric image, wait for at least 10 minutes for services to come up. If the services do not come up, the screen can display a message such as the following:
    7.6.0-mapr-devdocker-container % ./datafabric_container_setup.sh
Please enter the local sudo password for <username>
Password:

latest: Pulling from maprtech/edf-seed-container
Digest: sha256:052f461d98b1d0b8251cd47bab71b42103e61aaaa33d31335d3ca60182f4a87e
Status: Image is up to date for maprtech/edf-seed-container:latest
docker.io/maprtech/edf-seed-container:latest
Developer Sandbox Container b4be66858760 is running..
services required for Ezmeral Data fabric are coming up
services required for Ezmeral Data fabric are coming up
services required for Ezmeral Data fabric are coming up
services required for Ezmeral Data fabric are coming up
services required for Ezmeral Data fabric are coming up
services required for Ezmeral Data fabric are coming up
services required for Ezmeral Data fabric are coming up
services required for Ezmeral Data fabric are coming up
services required for Ezmeral Data fabric are coming up
services required for Ezmeral Data fabric are coming up
services required for Ezmeral Data fabric are coming up

services required for Ezmeral Data fabric are coming up

services didnt come up in stipulated 10 mins time 
please login to the container using ssh root@localhost -p 2222 with mapr as password and check further 
For documentation on steps to debug, see https://docs.ezmeral.hpe.com/datafabric/home/installation/troubleshooting_seed_node_installation.html 
once all services are up fabric UI is available at https://<hostname>:8443/app/dfui and fabrics can be deployed from that page
  2. Sign in to the Docker container using mapr as the password:
    ssh root@localhost -p 2222
  3. Enter the jps command and check the output. Continue entering the jps command until the command shows the AdminApplication java process, which indicates that all the services are started:
    root@edf-installer:~# jps
71136 FsShell
71315 Jps
19349 AdminApplication
10024 WardenMain
14236 CLDB
13213 QuorumPeerMain
  4. If the services do not start, check that sufficient resources have been allocated to the seed node. See the "Seed Node Prerequisites" in Fabric Deployment Using a Seed Node. You might need to allocate more resources and retry installing the seed node.
  5. Check to see if the Warden and ZooKeeper services are up and running:
    systemctl status mapr-warden
systemctl status mapr-zookeeper
  6. If the services did not start within 10 minutes, check the following logs for errors or exceptions. If the logs contain errors or exceptions, contact HPE Support. If there are no errors or exceptions, start the services (see step 7).
    • /opt/mapr/logs/cldb.log
    • /opt/mapr/logs/configure.log
    • /opt/mapr/logs/warden.log
    • /opt/mapr/apiserver/logs/apiserver.log
    • /opt/mapr/zookeeper/zookeeper-3.5.6/logs/zookeeper.log
  7. If the Warden and ZooKeeper services are not up and running, try restarting the services manually:
    systemctl start mapr-zookeeper
systemctl start mapr-warden