incubator-cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Arya Goudarzi <agouda...@gaiaonline.com>
Subject Re: After loadbalance why does the size increase
Date Tue, 26 Oct 2010 23:09:33 GMT
You won't get even distribution. All it does is decommission and auto-bootstrap. http://wiki.apache.org/cassandra/Operations
. However, you will get closer to even distribution if you double the size of your cluster,
meaning if you had 3 nodes, you add 3 more nodes. In that case, each new node will assume
1/2 of an older node.

Right now the only way you could get blanced ring is by doing manual token moves. 1418 seams
to be addressing this issue. I will also look forward into it.

----- Original Message -----
From: "Joe Alex" <joe.m.alex@gmail.com>
To: user@cassandra.apache.org
Sent: Tuesday, October 26, 2010 2:38:07 PM
Subject: Re: After loadbalance why does the size increase

I did not. I did try "cleanup" and here is after that.
Was expecting an even distribution and total load to be approximately
same as before

also was reading this https://issues.apache.org/jira/browse/CASSANDRA-1418


Address       Status     Load          Range
           Ring

127314552263552317194896535803056965704
10.210.32.75  Up         684.52 MB
3261306534628430564231103090162243723      |<--|
10.210.32.92  Up         727.91 MB
38958085454962719903262501903937071633     |   |
10.210.32.93  Up         366.47 MB
81410091240220304547331026452103785021     |   |
10.210.32.74  Up         415.47 MB
127314552263552317194896535803056965704    |-->|


On Tue, Oct 26, 2010 at 5:28 PM, Arya Goudarzi <agoudarzi@gaiaonline.com> wrote:
> Do you perform nodetool cleanup after you loadbalance?
>
> ________________________________
> From: "Joe Alex" <joe.m.alex@gmail.com>
> To: cassandra-user@incubator.apache.org
> Sent: Tuesday, October 26, 2010 2:06:58 PM
> Subject: After loadbalance why does the size increase
>
> Hi,
>
> I have Cassandra 0.6.6 running on 4 nodes with RF=2. I have around 2
> million rows for a simple test.
> Can somebody tell me why does the size keep increasing after running
> "loadbalance"
>
> Address       Status     Load          Range
>            Ring
>
> 127384081359183520765545698844531833520
> 10.210.32.75  Up         458.08 MB
> 3261306534628430564231103090162243723      |<--|
> 10.210.32.92  Up         407.88 MB
> 38958085454962719903262501903937071633     |   |
> 10.210.32.93  Up         350.51 MB
> 81410091240220304547331026452103785021     |   |
> 10.210.32.74  Up         498.58 MB
> 127384081359183520765545698844531833520    |-->|
>
>
> Address       Status     Load          Range
>            Ring
>
> 127314552263552317194896535803056965704
> 10.210.32.75  Up         684.52 MB
> 3261306534628430564231103090162243723      |<--|
> 10.210.32.92  Up         624.67 MB
> 38958085454962719903262501903937071633     |   |
> 10.210.32.93  Up         350.51 MB
> 81410091240220304547331026452103785021     |   |
> 10.210.32.74  Up         924.65 MB
> 127314552263552317194896535803056965704    |-->|
>

Mime
View raw message