increasing replication level in live DC

[Chris Wirt <ch...@struq.com>]

Hi,

I want to increase the replication level in one of our datacenters from 1
to 2. This DC is live, so thought I'd do the right thing and email users,
instead of cowboying it.

It is a small DC with just 3 nodes, tonnes of space and quite strict
latency requirements. We are running 1.2.4 with vnodes and whole load of
tweaks to make reads as fast as possible.

What is the recommended procedure to achieve this with minimal affect on
performance?
and
What is the recommended procedure to achieve this as quickly as possible?


Thanks,
Chris

Re: increasing replication level in live DC

[aaron morton <aa...@thelastpickle.com>]

background http://wiki.apache.org/cassandra/Operations#Replication

Try:
* Assuming you are on CL one now, there should be little impact. 
* to reduce impact set read repair chance on the CF's to 0
* check the stream throughput and compaction throughput in yaml, assume you have
* increase the RF
* run repair -pr on one node at a time. 
* set read repair chance to 0.1 or keep it disabled if you want to
* change your CL

Hope that helps. 

-----------------
Aaron Morton
Cassandra Consultant
New Zealand

@aaronmorton
http://www.thelastpickle.com

On 20/07/2013, at 8:34 PM, Chris Wirt <ch...@struq.com> wrote:

> Hi,
> 
> I want to increase the replication level in one of our datacenters from 1 to 2. This DC is live, so thought I'd do the right thing and email users, instead of cowboying it.
> 
> It is a small DC with just 3 nodes, tonnes of space and quite strict latency requirements. We are running 1.2.4 with vnodes and whole load of tweaks to make reads as fast as possible.
> 
> What is the recommended procedure to achieve this with minimal affect on performance?
> and
> What is the recommended procedure to achieve this as quickly as possible? 
> 
> 
> Thanks,
> Chris