Release 0.6.0 (Jan. 19, 2017)¶
- 90+ issues resolved
- NiFi clustering support. Ability to cluster NiFi with Kylo.
- Streaming enhancements. New streaming UI plots and higher throughput performance between Kylo and NiFi. Ability to specify a feed as a streaming type to optimize display.
- Improved schema manipulation. Ability for feeds and target Hive tables to diverge (for example: drop fields, rename fields, and change data types for fields the exist in raw files regardless of raw type).
- Alert persistence. Ability for alert panel to store alerts (and clear) including and APIs for plugging in custom alert responder and incorporate SLA alerts.
- Configurable data profiling. Profiled columns can be toggled to avoid excessive Spark processing.
- Tags in search. Ability to search tags in global search.
- Legacy NiFi version cleanup. Deletes retired version of NiFi feed flows.
- Avro format option for database fetch. GetTableData processor has been updated to allow writing rows in Avro format and to allow setting a custom column delimiter when the output type is a delimited text file.
- Feed file upload. Ability to upload a file directly to a feed and have it trigger immediately (for feeds using filesystem).
- Tutorials. New admin tutorial videos.
- JMS topics switch to queues in order to support NiFi clustering. Check your ActiveMQ Topics page (http://localhost:8161/admin/topics.jsp) to ensure that there are no pending messages before shutting down NiFi. The number of enqueued and dequeued messages should be the same.
- Apache NiFi versions 0.6 and 0.7 are no longer supported. Some features may continue to function normally but haven’t been properly tested. These will stop working in future releases. Upgrading to the latest version of Apache NiFi is recommended.
- (for VirtualBox sandbox upgrades) The default password for MySQL has changed. If you are using default config files deployed via RPM, modify your MySQL password to match or alter the configuration files.
Upgrade Instructions from v0.5.0¶
Build or download the RPM:
- Shut down NiFi:
service nifi stop
- Run the following to uninstall the RPM:
- Install the new RPM:
rpm –ivh <RPM_FILE>"
- Update /opt/nifi/current/conf/nifi.properties file and change it to use the default PersistentProvenanceRepository:
- Execute the database upgrade script:
/opt/thinkbig/setup/sql/mysql/thinkbig/0.6.0/update.sh localhost root <password or blank>
- Create the “/opt/nifi/activemq” folder and copy the jars:
$ mkdir /opt/nifi/activemq $ cp /opt/thinkbig/setup/nifi/activemq/*.jar /opt/nifi/activemq $ chown -R nifi /opt/nifi/activemq/
- Add a service account for thinkbig application to nifi group. (This will allow Kylo to upload files to the dropzone location defined in NiFi). This step will differ per operating system. Note also that these may differ depending on how the service accounts where created.
$ sudo usermod -a -G nifi thinkbig
All dropzone locations must allow the thinkbig service account to write.
- Start NiFi: (wait to start)
service nifi start
If errors occur, try removing the transient provenance data: rm -fR /PATH/TO/NIFI/provenance_repository/.
- Update, using your custom configuration, the configuration files at:
/opt/thinkbig/thinkbig-ui/conf/ /opt/thinkbig/thinkbig-services/conf/ /opt/thinkbig/thinkbig-spark-shell/conf/
A backup of the previous version’s configuration is available from /opt/thinkbig/bkup-config/.
- If using NiFi v0.7 or earlier, modify /opt/thinkbig/thinkbig-services/conf/application.properties by changing spring.profiles.active from nifi-v1 to nifi-v0.
- Start thinkbig apps:
- Update the re-usable standard-ingest template, index_schema_service, and the index_text_service.
- The standard-ingest template can be updated through the templates page. (/opt/thinkbig/setup/data/templates/nifi-1.0/) The upgrade will:
- Add “json field policy file” path as one of the parameters to Profiler processor to support selective column profiling. See “Configurable data profiling” in highlights.
- Add feed field specification to support UI ability to modify feeds. See “Improved schema manipulation” in highlights above.
- Adds shared library path to activemq libraries required going forward.
- The index_schema_service and index_text_service templates are feed templates and should be updated through the feeds page. (/opt/thinkbig/setup/data/feeds/nifi-1.0/.
- Go to the feeds page.
- Click the Plus icon.
- Click on the “import from file” link.
- Choose one of the Elasticsearch templates and check the overwrite box.
- A ReportingTask is now used for communication between NiFi and Operations Manager. In order to see Jobs and Steps in Ops Manager you will need to configure this following these instructions: