You can update the configuration (remove single disk that you don't need)  and perform a restart on one datanode at a time. Ensure that there are no under replicated blocks (namenode UI will show that)  in between datanode restarts. I found this to be safest approach. SDL ... View more

Can you not install hdfs client (just binaries) on the remote server and run a cron job to get hdfs data you want locally? That way you don't need to cache data locally and then rsync/scp to remote server?   SDL ... View more

Thanks Smarak,   In that case, how can we update the default configuration? I was under impression that once I upload my solrconfig.xml for my collection to zookeeper that was it. Each of my collections, ranger_audits, hadoop_logs and audit_logs has different values. Does that mean, I need to update default configuration for each of these separately (3 edits)? and potentially many more e.g atlas collections etc sdl ... View more

Thanks Smarak,   I did follow the the link and especially the section Old Installation; Solr Audits = enabled It worked as I could see older indexes were deleted. I could see disk space went down. However, for whatever reason, the ttl field reverted back next day. Not sure why/how? Also the autoDeletePeriodSeconds = 86400 never worked, and I am suspecting that we do stop the instances nightly. I just wanted to confirm that.   sdl ... View more