Hi Peter,
Thank you for helping me out!!!
I went through the provided link and it has helped me a lot. Now i do get
that there will be a intra worker messaging and another messaging system
for pulling data into storm.
Can you also tell me whether i need to have specific knowledge about the
Once per supervisor / worker, I should say.
Best regards,
Joel
2014-07-01 14:38 GMT+02:00 Joel Samuelsson samuelsson.j...@gmail.com:
We are using storm bolts written in python. In the bolt we have a
dependency that takes a lot of time to import. Is there a way to make the
import only once
Hi,
Well that depends on how you pull data into your storm cluster. If you are
pulling it from a queue, you will need to know how the queuing system
allows you to retrieve the data. You may have standards such as JMS or even
specific implementations as in kafka. Some people even pull data from
We use ActiveMQ with storm in production with a patched version of the JMS
Trident Spout.
On Wed, Jul 2, 2014 at 11:27 AM, Richards Peter hbkricha...@gmail.com
wrote:
Hi,
Well that depends on how you pull data into your storm cluster. If you are
pulling it from a queue, you will need to
actually I think this is a non-issue,
given the field exists in the stream already, I should be able to access it
right ?
On Wed, Jul 2, 2014 at 10:27 AM, Raphael Hsieh raffihs...@gmail.com wrote:
From my understanding, if I implement my own state factory to use in
PersistentAggregate, the
Yes. GroupingBy simply allows you to select out of the existing fields you
have created within your TridentTopology. You still maintain access to all
the fields created within the stream that is being groupedBy.
On Wed, Jul 2, 2014 at 1:44 PM, Raphael Hsieh raffihs...@gmail.com wrote:
actually
My hunch would be that you're anchoring your tuples and without acker tasks
these tuple trees are never destroyed. So all the tuples from each batch
are kept in memory until you run out of memory, i.e., a very, very fast
memory leak.
If you disable ackers, you should probably not anchor your
I couldn't find a reference to it, except for this old ticket:
But I was trying to use Clojure 1.6 with my local storm setup (and the
great Marceline https://github.com/yieldbot/marceline lib), and kept
getting the error of:
WARNING: some? already refers to: #'clojure.core/some? in namespace:
Whoops, forgot the ticket I was referring to:
https://issues.apache.org/jira/browse/STORM-265
On Thu, Jul 3, 2014 at 9:32 AM, Mark Mandel mark.man...@gmail.com wrote:
I couldn't find a reference to it, except for this old ticket:
But I was trying to use Clojure 1.6 with my local storm setup
Hi,
I really not remove anchor ack when I run storm in the unreliable way by
setNumAckers to 0.
I think it's simple by changing config to be between reliable and unreliable
storm. I'll try it later. Thank you
- Reply message -
寄件者： Cody A. Ray cody.a@gmail.com
收件者：
I believe the below is a stub you can flesh out.
def post_initialize( self ):
flesh out
Please note that I can't recommend Storm and Python; language layers begin
with Java - Jruby - Python, and the text pipe in between Jruby and Python
makes it really hard to recover from. I recommend
11 matches
Mail list logo
