Update documentation for restore and other cool features

[chrislovecnm]

Hey

I will probably knock this out myself, but this awesome tool may be lacking some documentation ... #justsayin

For instance:

cassandra-snapshotter -v --aws-access-key-id=redacted \ --aws-secret-access-key=redacted \ --s3-bucket-name=your_buck_name --s3-bucket-region=us-west-1 --s3-ssenc \ --s3-base-path=your_base_path restore --keyspace=your_keyspace --target-hosts=cassandra01,cassandra02

Also add more details what happens during a restore, for instance the program is going to download the snapshot out of the s3 repo to the machine you ran the command on, and then run sstableloader. You need disk and sstableloader. Also what are the details on prepping the cluster? Do you need the schema to exists?

I am thinking to provide examples of various commands, also more details on getting snapshots running properly. I am guessing that you have to list the keyspaces to get snapshotting running correctly.

Here is an example of a list command:

cassandra-snapshotter -v --aws-access-key-id=redacted \ --aws-secret-access-key=redacted --s3-bucket-name=your_bucket_name \ --s3-bucket-region=us-west-1 --s3-ssenc --s3-base-path=your_base_path list

Also we may want to link to https://help.ubuntu.com/community/SSH/OpenSSH/Keys for instructions on creating ssh shared keys.

The topics areas that I see need some TLC are:

• restore
• ssh keys
• list
• other command line switches that are not documented

Thoughts? Comments?

[chrislovecnm]

I have this work started in my help_options branch. Doing more testing before I pull.

[tbarbugli]

Documentation can be improved indeed; the reason why restore command is not documented is because I did not tested on production environments and because I did not try it at all after compression was introduced (so it might actually do nothing at this moment). Unless this changes and someone has time to test and fix the restore feature I would like to keep it hidden from users.

[arikfr]

Actually, it most likely won't work with the new versions that use compression. Unfortunately, I didn't have the time to update it to work with the compressed version. That's why we still use an older release of cassandra-snapshotter internally.

[tbarbugli]

adding lzop decompress should be really easy; what happens now is that the file object that is sent to s3 via boto is a generator consuming a pipe (eg. cat file | lzop).

reversing this on restore time should be quite easy.

in the coming two weeks i have no time for coding or testing but i can give pointer or do code review if needed.

Tommaso

On Saturday, 12 July 2014, Arik Fraimovich notifications@github.com wrote:

Actually, it most likely won't work with the new versions that use compression. Unfortunately, I didn't have the time to update it to work with the compressed version. That's why we still use an older release of cassandra-snapshotter internally.

— Reply to this email directly or view it on GitHub https://github.com/tbarbugli/cassandra_snapshotter/issues/27#issuecomment-48804100 .

sent from iphone (sorry for the typos)

[scalp42]

@tbarbugli @arikfr is there a branch somewhere with the restore working for the compressed backups?

[chrislovecnm]

Been open too long ... closing