moving all data to new cluster ?

["Langston, Jim" <Ji...@compuware.com>]

Hi all,

I have built a new 4 node cluster and would like to move the data
from the current 2 node cluster to the new cluster. What would be
the best way to move the data and utilize it on the new cluster. I
have looked at snapshot and also just copying the entire tree from
the old cluster to the new cluster. Not sure what the best practice
would be. I'm testing that process now in preparation for moving the
current data (production) to the new larger ring (systems are bigger
and more memory) and decommission the older ring (smaller systems,
less memory).

Thanks,

Jim

Re: moving all data to new cluster ?

[Robert Coli <rc...@eventbrite.com>]

Oh, I feel compelled to emphasize that you MUST avoid name collision when
doing this operation. Be CAREFUL to not overwrite files as a result of name
collision.

=Rob

Re: moving all data to new cluster ?

[Robert Coli <rc...@eventbrite.com>]

On Wed, Sep 4, 2013 at 1:01 PM, Langston, Jim <Ji...@compuware.com>wrote:

>  Thanks for the link Rob, but I did try earlier to
> copy the SSTables over and then to refresh them,
> but this is a brand new cluster and the error I got
> back indicated that the keyspace didn't exist, and
> then figured I needed to copy everything over in
> the data directory.
>

Just dump the schema from the old cluster and then load it on the new
cluster.

1) cassandra-cli "show schema"
2) create CFs one at a time (waiting in between for schema agreement)
3) nodetool refresh

=Rob

Re: moving all data to new cluster ?

["Langston, Jim" <Ji...@compuware.com>]

Thanks for the link Rob, but I did try earlier to
copy the SSTables over and then to refresh them,
but this is a brand new cluster and the error I got
back indicated that the keyspace didn't exist, and
then figured I needed to copy everything over in
the data directory.


Jim

From: Robert Coli <rc...@eventbrite.com>>
Reply-To: <us...@cassandra.apache.org>>
Date: Wed, 4 Sep 2013 12:44:12 -0700
To: "user@cassandra.apache.org<ma...@cassandra.apache.org>" <us...@cassandra.apache.org>>
Subject: Re: moving all data to new cluster ?

On Wed, Sep 4, 2013 at 12:38 PM, Langston, Jim <Ji...@compuware.com>> wrote:
I have built a new 4 node cluster and would like to move the data
from the current 2 node cluster to the new cluster. What would be
the best way to move the data and utilize it on the new cluster. I
have looked at snapshot and also just copying the entire tree from
the old cluster to the new cluster. Not sure what the best practice
would be. I'm testing that process now in preparation for moving the
current data (production) to the new larger ring (systems are bigger
and more memory) and decommission the older ring (smaller systems,
less memory).

http://www.palominodb.com/blog/2012/09/25/bulk-loading-options-cassandra

In your case, I would just copy all sstables to all target nodes and run cleanup.

=Rob

Re: moving all data to new cluster ?

[Robert Coli <rc...@eventbrite.com>]

On Wed, Sep 4, 2013 at 12:38 PM, Langston, Jim
<Ji...@compuware.com>wrote:

>  I have built a new 4 node cluster and would like to move the data
> from the current 2 node cluster to the new cluster. What would be
> the best way to move the data and utilize it on the new cluster. I
> have looked at snapshot and also just copying the entire tree from
> the old cluster to the new cluster. Not sure what the best practice
> would be. I'm testing that process now in preparation for moving the
> current data (production) to the new larger ring (systems are bigger
> and more memory) and decommission the older ring (smaller systems,
> less memory).
>

http://www.palominodb.com/blog/2012/09/25/bulk-loading-options-cassandra

In your case, I would just copy all sstables to all target nodes and run
cleanup.

=Rob