Closed tbarbugli closed 8 years ago
Hi! I really want Cassandra 2.1 support and going to implement it. Could you direct me where should I start?
afaik the main changes are related to where snapshots are stored (different naming convention). so you probably only need to change the part where the upload manifest is created. https://github.com/tbarbugli/cassandra_snapshotter/blob/master/cassandra_snapshotter/agent.py#L138 is where the list of files to upload (based on data_dir and snapshot name) is created.
What should be the signs of failure with Cassandra 2.1? I'm asking because I've tried cassandra_snapshotter from PyPI and it's backuped and listed correctly (at first glance).
well the absence of snapshot files on S3 for once
Well, it seems to fail when a table is specified, but works correctly with keyspaces. I believe it's due to table name postfix introduced in 2.1: glob.glob
doesn't catch it.
I think I know how to fix it. Is there an accepted way to distinguish Cassandra versions in the agent?
There is no cassandra version detected, I think the best would be to add a command line option to use cassandra 2.1 dir structure (unless you know a reliable to add version discovery). ᐧ
On Tue, Jun 16, 2015 at 10:31 PM, Ivan Yurchenko notifications@github.com wrote:
Well, it seems to fail when a table is specified, but works correctly with keyspaces. I believe it's due to table name postfix introduced in 2.1: glob.glob doesn't catch it.
I think I know how to fix it. Is there an accepted way to distinguish Cassandra versions in the agent?
— Reply to this email directly or view it on GitHub https://github.com/tbarbugli/cassandra_snapshotter/issues/23#issuecomment-112559966 .
data directory structure changed