nifi-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kevin Verhoeven <Kevin.Verhoe...@ds-iq.com>
Subject RE: setting processor concurrency based on the development/production environment
Date Fri, 02 Mar 2018 22:54:38 GMT
I am also interested in modifying concurrency based on dev/prod environment.

Do you know when 1.6 will be released? Is there a published roadmap for the releases?

Thanks!

Kevin

From: Ed B [mailto:bdesert@gmail.com]
Sent: Friday, March 2, 2018 10:23 AM
To: users@nifi.apache.org
Subject: Re: setting processor concurrency based on the development/production environment

Boris,
as the WA - I believe that after deploying the flow on another env, you can run post-deployment
scripts.
You could use PUT rest API to update processors by ID.
[image.png]



On Fri, Mar 2, 2018 at 12:04 PM Andrew Grande <aperepel@gmail.com<mailto:aperepel@gmail.com>>
wrote:
There are 2 efforts, with somewhat different focus. You are already aware of the community-driven
nipyapi, but there's also an official module Ientioned before, will be included with the 1.6
release.

Andrew

On Fri, Mar 2, 2018, 8:59 AM Boris Tyukin <boris@boristyukin.com<mailto:boris@boristyukin.com>>
wrote:
Hi Andrew,

thanks for the idea. I've been playing with nipyapi recently so might give this a try.

Thanks

On Thu, Mar 1, 2018 at 7:32 PM, Andrew Grande <aperepel@gmail.com<mailto:aperepel@gmail.com>>
wrote:
Boris,

Here's an idea youncould explore _today_.

Assume your dev and prod flows live in different bucket/registry instance. Given that you
are trying out NiFi 1.6, you should be able to extract the versioned flow from DEV and process
it to change the concurrency level for PROD before committing it to the prod registry instance.
Any script which understands json would do. nifi-toolkit-cli will take care of extracting
and moving flow versions.

It's not ideal (yes, would like concurrency to be a customizable flow var), and it assumes
an explicit process to promote between environments, but technically it is possible already.
The user experience can be improved in the future.

Andrew

On Thu, Mar 1, 2018, 1:52 PM Kevin Doran <kdoran@apache.org<mailto:kdoran@apache.org>>
wrote:
I think you could put it under either project. Ultimately, if we go with that approach, most
(all?) of the logic/enhancement would be in the NiFi code base during save version / import
flow / change version operations, so probably best to create it there.

Glad you are finding NiFi useful.

Cheers,
Kevin

From: Boris Tyukin <boris@boristyukin.com<mailto:boris@boristyukin.com>>
Reply-To: <users@nifi.apache.org<mailto:users@nifi.apache.org>>
Date: Thursday, March 1, 2018 at 13:44
To: <users@nifi.apache.org<mailto:users@nifi.apache.org>>
Subject: Re: setting processor concurrency based on the development/production environment

thanks Bryan and Kevin. I will be happy to open a jira - would it be a NiFi jira or NiFi registry?

I like the approach that Bryan suggested.

I guess for now I will just color code the processors that need to be changed in production.

P.S. I really, really like where NiFi is going...I've looked at StreamSets and Cask, but for
my purposes, I was looking for a tool when I can process various tables without creating a
flow per table. I was able to create a very simple flow in NiFi, that will handle 25 tables.
My next project is to handle 600 tables in near real-time. I just could see how I would do
that with StreamSets or Cask, when you have to create a pipeline per table. I was only being
able to do something similar with Apache Airflow, but airflow cannot do things in near real-time.
The concept of FlowFiles with attributes is a genius idea, and I am blown away with all the
possibilities to extend the functionality of NiFi with custom processors and Groovy scripts.
Awesome job, guys.

On Thu, Mar 1, 2018 at 1:29 PM, Kevin Doran <kdoran@apache.org<mailto:kdoran@apache.org>>
wrote:
Hi Boris,

Good point regarding concurrent tasks; thanks for sharing!

This is a great candidate for something that one should be able to create environment-specific
values for, as Bryan suggests. I agree we should create a NiFi JIRA to track this enhancement.

Thanks,
Kevin

´╗┐On 3/1/18, 11:44, "Bryan Bende" <bbende@gmail.com<mailto:bbende@gmail.com>>
wrote:

    Hello,

    Glad you are having success with NiFi + NiFi Registry!

    You brought up an interesting point about the concurrent tasks...

    I think we may want to consider making the concurrent tasks work
    similar to variables, in that we capture the concurrent tasks that the
    flow was developed with and would use it initially, but then if you
    have modified this value in the target environment it would not
    trigger a local change and would be retained across upgrades so that
    you don't have to reset it.

    For now you could probably always leave the versioned flow with the
    lower value of 2, then once you are in prod you bump it to 4 until the
    next upgrade is available, you then revert the local changes, do the
    upgrade, and put it back to 4, but its not ideal because it shows a
    local change the entire time.

    I don't think there is much you can do differently right now, but I
    think this is a valid case to create a JIRA for.

    Thanks,

    Bryan

    On Thu, Mar 1, 2018 at 11:29 AM, Boris Tyukin <boris@boristyukin.com<mailto:boris@boristyukin.com>>
wrote:
    > Hello NiFi community,
    >
    > started using NiFi recently and fell in love with it! We run 1.6 NiFi alone
    > with new NiFi registry and I am trying to figure out how to promote NiFi
    > flow, created in VM environment to our cluster.
    >
    > One of the things is "Concurrent Tasks" processor parameter. I bump it to 2
    > or 4 for some processors in my flow, when I develop / test it in VM.
    >
    > Then we deploy this flow to a beefy cluster node (with 48 cores) and want to
    > change concurrency to let's say 8 or 10 or 12 for some processors.
    >
    > Then I work on a new version/make some changes in my VM, and need to be more
    > shy with concurrency so set it back to 2 or 4.
    >
    > Then the story repeats...
    >
    > Is there a better way than to manually set this parameter? I do not believe
    > I can use a variable there and have to type the actual number of tasks.
    >
    >
    > Thanks
    > Boris



Mime
View raw message