Member since
11-20-2019
6
Posts
0
Kudos Received
0
Solutions
05-07-2023
02:27 PM
As per the OP's response if the data isn't well distributed along the partitioned column you will end up having some very large partitions while others will be very small. Writing into a single large partition can lead Kudu to fail. If your partitioned column is skewed aim for redesigning your table partitioning. Final note: As per Kudu's documentation (Apache Kudu - Apache Kudu Schema Design) Typically the primary key columns are used as the columns to hash, but as with range partitioning, any subset of the primary key columns can be used.
... View more
04-15-2020
07:54 AM
2 Kudos
@memad If your GetFile processor is consuming files before they have finished writing there are a few changes that may help: 1. How are files being written in to the directory? The default "File Filter" will ignore files that start with a ".". If it is possible to change how files are being written to the directory, that will solve your issue through a file filter. For example.... writing new files to directory as ".<filename>" and upon successful write does a rename to remove the dot (this is how ssh works). But you can of course setup any file filter that works for you, 2. Assuming the process that is writing files to the directory is always updating the timestamp on the file, you can use the "Minimum File Age" property to prevent the GetFile from consuming a file until the last modified timestamp in the file has not updated for the configured amount of time. This works in most cases, except when there may be long pauses in the write process that exceeds the configured Min File Age time. Hope this helps, Matt
... View more