anic.the logic is
> here:when write/get too large data to/from zk,it will give too much
> pressure on the server, sometimes cause the GC stop in the leader, follwers
> cannot ping the leader within the bounded time, then start to re-election.
> I guess,haha:D
> >
> > - Origin
ometimes cause the GC stop in the leader, follwers cannot ping the
> leader within the bounded time, then start to re-election. I guess,haha:D
>
> - Original Message -
> From: Asim Zafir
> To: maoling199210...@sina.com
> Cc: dev
> Subject: Re: Re: Jute buffer size
sometimes cause the GC stop in the leader, follwers cannot ping
the leader within the bounded time, then start to re-election. I guess,haha:D
- Original Message -
From: Asim Zafir
To: maoling199210...@sina.com
Cc: dev
Subject: Re: Re: Jute buffer size increasing.
Date: 2019-03-11 21:14
rver
> and client.
> - for the "jmap -histo:live" you had provided, you have created almost
> 103842 zk-nodes,and their memory usage is less than 10 MB, so don't worry
> about this.
>
> - Original Message -
> From: Asim Zafir
> To: Jordan Zimmerman
>
"jmap
-histo:live" you had provided, you have created almost 103842 zk-nodes,and
their memory usage is less than 10 MB, so don't worry about this.
- Original Message -
From: Asim Zafir
To: Jordan Zimmerman
Cc: Zookeeper
Subject: Re: Jute buffer size increasing.
Date: 2019-03-
thanks and very much appreciate your feedback.
here the system that is using zookeeper is HBASE and you are right its a
manual setting. we had it 1mb, then increase to 20mb, 32mb and now 128 mb..
what is the impact of having higher jute buffer size and more (higher count
of org.apache.zookeeper.da
The Jute buffer size is a manual setting, so it wouldn't increase by itself. If
you find that you have to keep increasing it's due to a few possibilities:
ZNodes with lots of children (with long names). Any 1 ZooKeeper API call is
limited by jute max buffer. So, if a call to getChildren() where
+ ZK dev community. Please assist.
On Fri, Mar 8, 2019 at 4:10 PM Asim Zafir wrote:
> Hi Jordon,
>
> We are seeing constance increase in jute buffer size on our zookeeper
> instance. right now it is set to 128. We are primarily using zookeeper for
> HBase cluster. I want to see what is contributi