By default, I have these:
hadoop.proxyuser.hive.hosts=*
hadoop.proxyuser.hive.groups=*
Also, I have impersonation enabled. These settings should have made the
compaction to run without issues, but does not. What am I missing?
On Wed, Nov 16, 2016 at 6:00 PM, Manoj Murumkar
wrote:
> Okay, made
Okay, made the cleaner work as well :)
When the files are owned by a user that is different from the user that
runs metastore ("hive"), this doesn't work. Just to test this, I changed
ownership of the files for the table to "hive" and ran a major compaction
and it does the job.
drwxrwxrwt - hiv
Further update:
I see this:
+---+--+++-+---++--+
| dbname | tabname | partname | type |state|
workerid | starttime|
+---+--+++-+--
Quick update:
After each compaction, files under base directory (for the buckets) have
latest data. However, I am expecting to see all delta files (and
directories) gone, as they should be merged in the base directory.
Otherwise, we'll start seeing too many small files on HDFS which is a
problem.
Hi,
We are trying to implement transaction feature in hive. I created following
table:
++--+
|
createtab_stmt |
+
