basho / cuttlefish

never lose your childlike sense of wonder baby cuttlefish, promise me?
Apache License 2.0
205 stars 124 forks source link

effective config doesn't include advanced.config mods #127

Closed joedevivo closed 10 years ago

joedevivo commented 10 years ago

This is a tricky one, I'd like to get one of the following solutions in for riak's 2.0 release:

Option 1

simply cat the advanced.config at the end of the output of the effective conf. It will at least indicate to the user that these values are detected and being used.

Option 2

Since we know which app.config values a cuttlefish variable will effect, we could include some kind of indicator that it is overridden in advanced.config.

e.g.

some.setting.a = on
## some.setting.b is overriden in advanced.config's 'erlang_library.variable'
## some.setting.b = off

I think this option is more friendly to commandline querying of effective confs. grep etc...

joedevivo commented 10 years ago

Something needs to be done for 2.0 even if it's just a disclaimer. bottom line is that even if it can't be accurate, which I think it can be, it can't lie

joedevivo commented 10 years ago

riak config effective should also be tested for nodes configured with legacy app.config and vm.args, and at the very least print an error message that these legacy files are not supported by 'effective'

cmeiklejohn commented 10 years ago

The testing should be tagged with 'test' and 'resilience'.

joedevivo commented 10 years ago

@cmeiklejohn the testing of this is covered by #142

bookshelfdave commented 10 years ago

all eunit tests pass: https://gist.github.com/metadave/9896753

bookshelfdave commented 10 years ago

it looks like advanced.config vars that don't have a cuttlefish equiv (such as repl) won't show up in the output.

bookshelfdave commented 10 years ago

latest eunit output: https://gist.github.com/metadave/9901069

bookshelfdave commented 10 years ago

+1 52e963d

joedevivo commented 10 years ago

@borshop merge

cmeiklejohn commented 10 years ago

Super pedantic, but in the future I'd really like to see commit histories like this squash so there aren't a bunch of broken WIP intermediary commits in the history.

jrwest commented 10 years ago

+1 to that @cmeiklejohn