Backup implementation for Ensembles

[GoogleCodeExporter]

How do you scale a currently running ensemble from N clusters to N+1
cluster without losing data?

I had a ensemble running of 2 clusters (each with 1 master and 1
server)
I tried the following, but could not retain all the data (I had just 1
bucket):

1. Export backup of the bucket of both the servers.
2. Restart the servers using the update cluster.json files. So now I
have a ensemble of 3 cluster (the third cluster is also 1 master and 1
server)
3. Import the backup on both the servers.
4. When I try to GET all data, it shows 404 for some of the records.

I even tried cleaning the masters before joining the servers in the
new ensemble, but faced the same issue.

Original issue reported on code.google.com by rohi...@gmail.com on 10 Jun 2011 at 2:58

[GoogleCodeExporter]

Hey Rohit,

the bug should be fixed on trunk.
To execute a backup export/import you need to:
1) Execute a bucket export on one server for every ensemble cluster: i.e., if 
you have an ensemble with 2 clusters and 4 servers, just execute the export on 
two servers (one per cluster).
2) Import all previously exported backups: Terrastore will take care of 
properly routing the imported data even if ensemble has changed.

Any chances you can try it out?

Original comment by sergio.b...@gmail.com on 1 Jul 2011 at 4:29

• Changed state: Started

[GoogleCodeExporter]

Original comment by sergio.b...@gmail.com on 2 Jul 2011 at 3:45

• Changed state: Fixed
• Added labels: Type-Defect, Priority-High, Project-Terrastore, Milestone-0.8.2