adding the header 'X-Riak-Timeout=' will now cause the request to return a timeout error if it takes longer than the specified time interval. This works for DELETEs, PUTs, and GETs, but in the case of a PUT or a get, a timeout failure does not guarantee anything about the state of the cluster when it happens. The PUT or DELETE may or may not have succeeded; the correct course of action after the timeout is to retry the same action until a 200 is received.
expose timeouts via webmachine
- remove redundant default
- clarify error message
The associated riak_test is here: basho/riak_test#150
Too late for 1.3, but I like the general idea.
Couldn't make heads or tails of the travis output.
@evanmcc Not to delay this PR, but we should also add these to the appropriate PB messages.
@evanmcc Yes, rt:pbc(Node) will grant you a new client.