Mailing-List: contact user-help@flink.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@flink.apache.org
MIME-Version: 1.0
In-Reply-To: 
 <CANMXwW3G24RjLEnDc22h7RRKw302oMjt4SRJj-C=C=79CpvyyQ@mail.gmail.com>
References: 
 <CA+QjRbTogm91a4sfckNX4_bHDrLHNb3CsUkFmVmRewqi3CT2OQ@mail.gmail.com>
	<CANC1h_v2-X-HaySraNR1079TcxZdOnw_wNpFQNKRhv3kV0V4vw@mail.gmail.com>
	<CA+QjRbQDLw2jTFZXjfFgXT5SN0KOuRPcPj_7ZSYNQTzsWpTy9A@mail.gmail.com>
	<CAGr9p8DOJ9BnL8=2QjXC=54jVmmHDsUVgjhfT7DVtUXPvM2_yg@mail.gmail.com>
	<CANMXwW3G24RjLEnDc22h7RRKw302oMjt4SRJj-C=C=79CpvyyQ@mail.gmail.com>
Date: Mon, 4 Apr 2016 22:54:07 +0900
Message-ID: 
 <CA+QjRbTdLz7SfhoMuUrWLcZKFrX8pRAsMg+_G+1t5HoVHSqAYA@mail.gmail.com>
Subject: Re: scaling a flink streaming application on a single node
From: Shinhyung Yang <shinhyung.yang@gmail.com>
To: user@flink.apache.org
Content-Type: text/plain; charset=UTF-8

Dear Aljoscha and Ufuk,

Thank you for clarifying! Yes I'm running this wordcount application
on a 64-core machine with 120GB ram allocated for users.

> In that case, the amount of RAM you give to the TaskManager seems to low.
> Could you try re-running your experiments with:
> jobmanager.heap.mb: 5000
> taskmanager.heap.mb: 100000
>
> So, about 100 GB of RAM for the TaskManager.

Definitely I will try this! The result will be really interesting for
sure. In this case, am I still good to go with 64 task slots with a
single task manager?

Thank you.
With best regards,
Shinhyung Yang.