The Distributed portion is the agents downloading it from the CM server. Check the CM Agent logs for clues. The obvious issues will be resolution of the CM server address or ports being blocked. ... View more

Are sure they parquet files? I haven't used the parquet tools in a bit so I don't know the behavior of trying to read a non-parquet file with them. I could see it view it as not existing if that is the case and it would also be why you can see the data through the table. How was this data added to HDFS/table? Is this a Hive or Impala table? ... View more

Try to read the file directly. There are other parquet tool options as well to validate the schema, etc. /opt/cloudera/parcels/CDH/lib/parquet/bin/parquet-tools cat /path/to/parquet.file ... View more

You could do that and create your own CA cert and then use that to sign your server certs. You should be able to copy the CM self-sign certificate to each hosts and configure that as well. Just the cert, PEM file, not the key. You will need to have this set up so that the Agents can talk to CM over TLS to move the keytabs around, if using the wizard. If you are not using the wizard you can still configure Kerberos and use it without going any further. ... View more

What is the output of 'show partitions <tablename>;'? I suspect that HMS does not know about the partitions. If you don't see any partitions in the output you can add them manually using one of the below commands. This will add all missing partition that it finds in HDFS. msck repair <tablename>; This will add one at a time: alter table <tablename> add partition (y=2017,m=02,d=23,h=23); ... View more

Download Firefox and configure. See if it works before we dig into Chrome. That way we will know that we just need to get Chrome to negotiate with the site. ... View more

Can you upload the full role logs for one or two of the services failing to start? ... View more

If you are talking about Worker nodes (datanode, nodemanager, impala daemon, etc.) and the resource allocation and configuration is identical, then in theory yes it should be a 2x improvement. Note: you will need to rebalance the data across the new nodes to see consistent improvement and not burnt out the old nodes. ... View more

I haven't done Spark History server specific but not had issues with others. It could be a bug. I'll see if I can reproduce. What CM/CDH versions? ... View more

How did you move it? Did you remove the role form the one host and then add it to the other in CM? ... View more