We were doing it as backup and super backup.
That's interesting though, thanks for the suggestion. I thought we
needed to have both.
So if the local is on ephemeral storage and we restart the node (eg
local is wiped), it'll just sync with S3, join the cluster, and go?
On Apr 19, 4:25 pm, Shay Banon shay.ba...@elasticsearch.com wrote:
You should also change under the data directory of each node the name of hte cluster. The format is: data/cluster_name/nodes/N/... .
Out of curiosity, why do you use both EBS for data location, and s3? If you use s3 gateway, then you can keep the data on the local instance disk, its already backed in s3.
If you want ot use EBS, then use local gateway, will be faster and use less resources (s3).
On Wednesday, April 20, 2011 at 1:35 AM, mr.gaffo wrote:
I've got a production cluster that I would like to mirror down to
development intact. I'd like to rename the cluster so I don't get
confused and things don't conflict. What would I need to do to do
this. My gateway is up on S3 and I have local files in EBS for the
nodes. I was going to copy the local files to the new cluster nodes
and the S3 files to the new bucket. What else do I need to do.