Reply
Highlighted
PNS
Explorer
Posts: 37
Registered: ‎05-20-2014
Accepted Solution

Morphlines and multithreading

Hi...

 

Does the single-threaded design of morphilnes mean that there can be concurrency issues by creating multiple morphlines from the same config file and running each of them in a separate thread?

 

For example, if one compiles 10 morphlines from the same config file and then run each of those in a separate thread, can that cause any thread safety issues?

 

Thanks.

Cloudera Employee
Posts: 144
Registered: ‎08-21-2013

Re: Morphlines and multithreading

bq. For example, if one compiles 10 morphlines from the same config file and then run each of those in a separate thread, can that cause any thread safety issues?

There?s no thread safety issue when doing that. (And there?s minimal contention as well, which improves throughput). In fact, that?s exactly what the hbase-indexer does in order to take advantage of multiple CPU cores, and that?s also what users of the Flume Morphline Solr Sink sometimes do.

Wolfgang.

PNS
Explorer
Posts: 37
Registered: ‎05-20-2014

Re: Morphlines and multithreading

I assumed so, but wanted to confirm that there is no state sharing "behind the scenes", especially in the Command building process.

 

Thanks for the confirmation.

Announcements
The Kite SDK is a collection of docs, sample code, APIs, and tools to make Hadoop application development faster. Learn more at http://kitesdk.org.