About mburgess

mburgess · ‎10-30-2018

If you're using a DistributedMapCacheServer, you can set a "Persistence Directory" and it will store the cache to disk, so it will be available on restart. DistributedMapCacheServers are started on each node in a cluster, but your DistributedMapCacheClientService provides the hostname and port to only one of them. They do not coordinate to keep the same data, so you will want all your clients (for put and get) to point to the same instance of the server. Since you really just want the information from Cassandra, please feel free to file a New Feature Jira to add a CassandraMapCacheServer, then you could fetch the data directly from Cassandra.

mburgess · ‎10-25-2018

Is "Translate Field Names" set to true in PutDatabaseRecord? If not try that. If so, try removing the "namespace":"dbo" from the schema. If that works, then it's probably a bug, as we should be trying to match on fullname but fall back to simple field names.

mburgess · ‎10-25-2018

Do you have multiple concurrent tasks running? If so, you might be running into this issue, and the workaround seems to be to use a single task. If not, are you importing any native (CPython) libraries? You won't be able to do that with Jython, but I thought it gave a different error message in that case.

mburgess · ‎10-24-2018

Even though it's managed by Hive, unless Hive is performing tasks as the Hive user (which here it is not), I believe Shu is right and you'll likely need an HDFS policy in Ranger to allow user nifi to access the hive warehouse.

mburgess · ‎10-22-2018

There is a mongo-java-driver you can use in either Groovy or Javascript (or any of the scripted languages in ExecuteScript), check my blog on how to use modules.

mburgess · ‎10-19-2018

Your "Database type" property is set to "Generic", try setting it to Oracle (for Oracle < 12) or Oracle 12+.

mburgess · ‎10-15-2018

I'm not sure if I fully understand how your downstream flow would work when there are < 9 flow files available to ExecuteScript, but here's something to try (note that you can use anything greater than 9 in place of the "10" below): def flowFiles = session.get(10) if(flowFiles.size() >= 9) { session.remove(flowFiles) } else { session.transfer(flowFiles, REL_SUCCESS) }

mburgess · ‎10-15-2018

What about if there are zero available? Are you scheduling this processor to run every few seconds, or the default (as fast as possible)? If the latter, you will likely want to check that there are > 0 and < 9 flow files available.

mburgess · ‎10-12-2018

The entire flow file content is one "row" to be inserted into the database? You should still be able to use this, as of NiFi 1.7.0 the JsonTreeReader should recognize "one JSON per line" rather than expecting an array.

mburgess · ‎10-12-2018

If the "mongo" library contains native Python (CPython) code or dependencies, you can't use them in ExecuteScript (which uses Jython not Python). See Matt C's answer for a workaround to use actual Python.

Online	Offline
Last Visited	‎01-16-2025 06:21 PM

Member Since	‎11-16-2015 02:21 PM
Last Visited	‎01-16-2025 06:21 PM
Posts	892
Kudos received	643

Cloudera Community

Re: Nifi Building error when creating a brand new ...

Re: Tuning PutHive3Streaming NiFi processor

Re: NiFi ExecuteScript - Able to add attributes to...

Re: NiFi - JOLT assign value to attribute from Jso...

Re: NiFi - ExecuteScript for getting max value of ...

Re: How to update nifi cache on nifi service resta...

Re: Apache NiFi PutDatabasRecord processor error: ...

Re: Executescript error

Re: putHive3Streaming Error

Re: NIFI -- How to import python libs for ExecuteS...

Re: NiFi + Oracle: ORA-00933: SQL command not prop...

Re: NiFi - Script count fileFlows Groovy

Re: NiFi - Script count fileFlows Groovy

Re: Nifi Processor : Insert data into CLOB column

Re: NIFI -- How to import python libs for ExecuteS...