Cannot kill topology

2014-11-16 Thread Oleg Tokarev
Hello, Problem is: - Started (storm jar ...) DRPC topology with ShellBolt (PHP) (ver 9.3.0-rc1) - After being started it works ok for us - After being killed there is a garbage left somewhere according logs (see logs below) What can it be? In what direction should i continue my investigat

Re: Storm's comparison towards Pipes-and-Filters-Pattern

2014-11-16 Thread Patrick Wiener
@Nathan: :) thanks for the quality comparison. Nice bottom line for a presentaton/conclusion. @all: thank you for your help. Awesome support in this community. > Am 17.11.2014 um 07:58 schrieb Vladi Feigin : > > Nathan, > > Liked ! <> > > >> On Sun, Nov 16, 2014 at 9:06 PM, Nathan Leun

Re: Storm's comparison towards Pipes-and-Filters-Pattern

2014-11-16 Thread Vladi Feigin
Nathan, Liked ! <> On Sun, Nov 16, 2014 at 9:06 PM, Nathan Leung wrote: > Also it supports cycles in the graph. It's like comparing a bicycle to an > airplane. > On Nov 16, 2014 2:03 PM, "Vladi Feigin" wrote: > >> Storm is much more sophisticated then just filter-pipe pattern. >> It provide

Shutting down long running bolts

2014-11-16 Thread Toby Hobson
I'm trying to shutdown a long running bolt (based on the sample ExclamationBolt) but it seems the cluster.killTopology() call is only interrupting one task and continuing to execute other tasks. To simplify things I am only running one bolt in the topology. This is the bolt code I am running (local

Re: Storm's comparison towards Pipes-and-Filters-Pattern

2014-11-16 Thread Nathan Leung
Also it supports cycles in the graph. It's like comparing a bicycle to an airplane. On Nov 16, 2014 2:03 PM, "Vladi Feigin" wrote: > Storm is much more sophisticated then just filter-pipe pattern. > It provides > 1. Reliability: guarantees that every spout tuple will be fully > processed. Actuall

Re: Storm's comparison towards Pipes-and-Filters-Pattern

2014-11-16 Thread Vladi Feigin
Storm is much more sophisticated then just filter-pipe pattern. It provides 1. Reliability: guarantees that every spout tuple will be fully processed. Actually it provides : at-most-once delivery(no ackers) , at-least-once delivery(ackers) and exactly-once (Trident) semantic for the message deliv

Re: Storm's comparison towards Pipes-and-Filters-Pattern

2014-11-16 Thread Patrick Wiener
So basically Storm’s core concept can be compared to pipes-and-filters-pattern BUT provides a more „user-friendly“ framework than e.g. a unix based pipes-and-filters processing. btw: I haven’t come across with TRIDENT yet. Just starting to dive deeper into Storm as a potential technology for a

STORM-426 - Upgrade to Clojure 1.6

2014-11-16 Thread Anuj Kumar
Hi, We have recently started facing an issue due to an upgrade to Clojure 1.6 for one of the component that we are using with the condition of aot being enabled. We can still workaround without aot and let it compile with Clojure 1.5 but Storm being able to run with Clojure 1.6 is what we are inte

Error-reporting format in Storm UI (0.9.0)

2014-11-16 Thread Exhausted
Hello all, In the Storm UI if I navigate to the page for a particular bolt there is a section at the bottom that shows errors associated with the bolt. The timestamps shown for the error messages are of a form like this: Fri, 07 Nov 2014 19:53:55 + The errors are displayed in sorted form,

Re: Max Spout Pending Ideal Number?

2014-11-16 Thread Vladi Feigin
hi, It's matter of fine tuning and depends on your topology , there is no one gold number. Start from 1K , monitor it. If you see that your topology supports higher throughput , increase it ... or if it's not, decrease it.. Vladi On Sat, Nov 15, 2014 at 9:27 AM, Nilesh Chhapru < nilesh.chha...