TIBCO Spotfire Data Science doesn't report status or results for my MapReduce job.

Rich Talbot

Created
October 08, 2021 23:28

Date Posted: 2018-01-23
Product: TIBCO Spotfire®

Problem:

Solution:

A job is submitted, and completed on the Hadoop cluster, but my results console does not show status or results.

Resource Manager Failover
The Resource Manager may fail over if High Availability (HA) is enabled.

Check which Resource Manager is active.
Verify the Data Source connection uses the active Resource Manager.

Websockets vs. LongPolling
TIBCO Spotfire Data Science results display requires communication from the Hadoop cluster to the TIBCO Spotfire Data Science server. By default, Websockets are used.

Check if my company's network has websockets enabled: http://websocketstest.com/
If websockets are not enabled, the other option is LongPoll.
1. Edit: $CHORUS_HOME/shared/ALPINE_DATA_REPOSITORY/configuration/alpine.conf
2. Add: alpine.ui.console.protocol=longpoll
3. Restart TIBCO Spotfire Data Science: $ chorus_control restart alpine

Bad Data Reporting
TIBCO Spotfire Data Science reports "bad data," i.e. missing data entries or data with unexpected type, to the results console. If the bad data reporting is delayed, make sure all Hadoop nodes are able to resolve the TIBCO Spotfire Data Science server hostname (eg nslookup <server_hostname>).

Using AWS:
1. Add an entry in the VPC security group to allow inbound access from the external IPs of all Hadoop nodes.
  - or -
2. Give the TIBCO Spotfire Data Science instance a hostname.
  1. Create an Amazon Route 53 DNS entry (Your laptop needs to have this DNS entry too).
    - or -
  2. Add the hostname / ip manually to /etc/hosts to your laptop and all Hadoop nodes.
Using on-premises servers:
1. DNS entry (Your laptop needs to have this DNS entry too).
  - or -
2. Add the hostname / ip manually to /etc/hosts to your laptop and all Hadoop nodes.