The --ipc flag is not configurable through YARN at this time. There is a ticket open for adding this feature, YARN-8447. No one is currently working on this, so it could be helpful for you to add a comment to the ticket describing your use case. ... View more

You can configure a restart policy for each component. Add a field like the following: "restart_policy": "ON_FAILURE" The available restart policies are ALWAYS, ON_FAILURE, and NEVER. The default policy is ALWAYS. ... View more

The accumulo shell will only work from a host where an Accumulo Client is installed. You can check in the Ambari UI to find out where clients are installed. ... View more

The client software does not need to be installed on all the nodes, only on the node(s) where client / CLI commands will be issued. In Slider's case, Slider jars will be uploaded in a tarball to HDFS from the client and YARN will localize the tarball for Slider's containers. ... View more

It should be possible to run Solr on YARN using Slider in HDP 2.4.2: https://github.com/hortonworks/slider-release/tree/HDP-2.4.2.0-tag/app-packages/solr Theoretically you would be able to launch a Solr app on YARN through the Slider View in Ambari, but I am not sure if anyone has tried this out for Solr. The Slider View for Ambari says it's marked as deprecated, so I am not sure of its status: https://docs.hortonworks.com/HDPDocuments/Ambari-2.2.2.18/bk_ambari-views/content/ch_using_slider_view.html ... View more

HBase can run on YARN or without YARN. MapReduce jobs always run on YARN, so whenever there is an HBase operation that performs a MapReduce, that will be run on YARN. ... View more

Slider uses its own Hadoop jars in the Slider lib directory. I suspect that the Hadoop client jars Slider is using are not compatible with the version of Hadoop running on the cluster. If you built Slider yourself, you could try rebuilding it with a hadoop.version that matches the version running on your cluster (you may need to adjust some other dependency versions in this case as well). Alternatively, you could try replacing the hadoop jars in the Slider lib directory, but if the versions are too far apart this will not work and you'll have to rebuild Slider. ... View more

I don't think the default hbase app package will allow you to get the region server ports through the REST API because Slider is not allocating those ports. The port is set to 0 in the hbase-site config, so hbase will pick a random port itself. If you wanted Slider to allocate the port instead, you might be able to change the hbase.regionserver.port property to the following in the appConfig: "site.hbase-site.hbase.regionserver.port": "${HBASE_REGIONSERVER.ALLOCATED_PORT}{PER_CONTAINER}", You would also have to add componentExports to the HBASE_REGIONSERVER component in the metainfo.xml file: <component> <name>HBASE_REGIONSERVER</name> <componentExports> <componentExport> <name>regionservers</name> <value>${THIS_HOST}:${site.hbase-site.hbase.regionserver.port}</value> </componentExport> </componentExports> ... You should then be able to access this information through /ws/v1/slider/publisher/slider/componentinstancedata ... View more

As described in the answer to this question: https://community.hortonworks.com/questions/35138/target-replicas-is-5-but-found-3-live-replicas.html Start the accumulo shell on a host where the Accumulo client is installed and run the following: config -t accumulo.root -s table.file.replication=3 config -t accumulo.metadata -s table.file.replication=3 You'll have to change these properties AND run the hdfs setrep command so that both new and existing files have their settings changed. ... View more