Riak Recap for 5/31

Mark Phillips mark at basho.com
Tue Jun 1 13:34:14 EDT 2010


Morning, Afternoon, Evening to all -

This Recap (as well as tomorrow's and Thursday's) is coming to you
live from the Open Source Bridge Conference in Portland, OR.  (For
anyone who is interested, there is a ustream happening here --->
http://is.gd/cymL5  - no promises on what will be broadcasted)

Without further delay, today's recap: a few great questions from #riak
and a pointer to a great Riak +node.js presentation.

Enjoy,

Mark

Community Manager
Basho Technologies
wiki.basho.com
twitter.com/pharkmillups

-----

Riak Recap for 5/31

1) Q --- How does M/R jobs behave if one or more nodes are unavailable
(due to net split, failure or such)? Does it simply skip processing
data in unavailable partitions or will it time out? (from chids via
#riak)

    A --- If a node fails before a map/reduce job begins processing
then data could be skipped if the failed node is the last node hosting
a given piece of data. Otherwise jobs should process normally unless
the node hosting the reduce phases fails. In that case, a timeout will
occur.

2) Q --- Does riak replicate the bucket information (e.g. list of keys
inside the bucket)? (from ddmmi via #riak

    A --- It's not replicated, it's gossiped along with ring state.

"Gossiping" in the Riak Glossary --->
http://wiki.basho.com/display/RIAK/Riak+Glossary#RiakGlossary-Gossiping

3) seancribbs and misaka had a brief exchange in #riak about single
node setups and durability that was just long enough to justify a
Gist.

Get it here ---> http://gist.github.com/421157

4) @frank06 posted his slides from the recent presentation he gave on
Riak and Node.js

Check them out here ---> http://www.slideshare.net/frank06/nodejs-riak



More information about the riak-users mailing list