We aren't in production currently. I was mostly asking if running LocalCluster 
in production is something that we should stay away from doing. It seems that 
this is the case, which makes sense.

From: [email protected] 
Subject: Re: In Process Zookeeper Log Clean Up

I am curious about your use case, Why are you running LocalCluster in 
production? 
You are missing out on distributed, high availability features when running in 
cluster mode.


On Fri, Dec 8, 2017 at 2:37 PM, Mitchell Rathbun (BLOOMBERG/ 731 LEX) 
<[email protected]> wrote:

When running Storm using LocalCluster, it starts an in process zookeeper 
server. Normally, when running a zookeeper server, it is up to the 
administrator to remove old logs so the directory specified by datalogdir 
doesn't fill up. The zookeeper config settings autopurge.snapRetainCount and 
autopurge.purgeInterval allow this to be done automatically. I am wondering if 
Storm automatically handles the clean up of these logs or provides a config 
file for the above settings to be changed from defaults. From looking through 
the code, it seems that it doesn't. I am guessing this is because local mode is 
for testing and isn't really meant to support long running topologies. Does 
running Storm through LocalCluster provide a way to prevent Zookeeper server 
logs from overflowing? And in general, is LocalCluster meant to be used for 
long running topologies in production, or just testing?


Reply via email to