Closed matthewvon closed 8 years ago
Note: in leveldb, the compression setting in LOG file will be 0 - disabled 1 - Snappy 2 - LZ4 (anything else is a severe bug!!)
After a quick discussion with @angrycub we'd rather see this as something like:
eleveldb.compression = on | off
eleveldb.compression.algorithm = snappy | lz4
Cuttlefish convention generally favors "simple" settings used in combination vs. single complex settings like what is proposed here. Additionally, we then end up with 2 Erlang settings (is compression on or off, and which algorithm should we use) as independent variables, rather than conflating the two into a single thing with on/off/true/false/snappy/lz4 options.
somebody know how to code that in cuttlefish? 'cuz the result is the same by the time it hits eleveldb (I believe from other options).
@matthewvon I can help you with the change to add algorithm
as a thing separate from the compression. I'm sure there are other examples of the above floating around we can use.
create jira issue
+1 cc0be0a
[~matthewv] Should this story for LZ4 compression for Riak move to "Needs Review" for someone to pick it up and review it? Didn't want to just move it forward without confirming with you.
_[posted via JIRA by Patricia Brewer]_
LZ4 options within riak.conf have passed review. Doug requests a new Jira ticket for updating Riak dependency lists with a new eleveldb tag. That tag will be ready tomorrow morning (Wed 8/9).
_[posted via JIRA by Matthew Von-Maszewski]_
This updates the eleveldb.schema file to accept "snappy" and "lz4" as independent options for the leveldb.compression variable. The default value remains "snappy", but end users now have the option to enable lz4 compression.