Hi Graham,
we usually mark breaking changes using the breaking label on github, you
can check out here the breaking changes included in
1.1: Issues · elastic/elasticsearch · GitHub
.
On the other hand, we try very hard to keep backwards compatibility between
minor versions to make sure rolling upgrades are possible, which is the
case between 1.0 and 1.1. Although there are some breaking changes, it's
important to read through them and understand on which level backwards
compatibility was broken (e.g. transport layer, rest layer etc.).
That said mixing nodes running different versions should be seen as a
temporary solution, e.g. till the rolling upgrade has completed, although
things work. What we effectively do on the transport layer is serialize and
de-serialize fields based on the versions of the current node and the node
we are talking to. That might mean writing/reading only the fields
supported by both nodes (aka downgrade to the lower version between the
two), which is something you dont wan't to happen unless you are doing
maintenance.
On Wednesday, March 26, 2014 11:50:43 AM UTC+1, Graham Tackley wrote:
The release notes for elasticsearch 1.1.0 don't say anything about
compatibility with 1.0 (or at least I didn't see it).
can I mix 1.0.1 and 1.1.0 in the same cluster, i.e. do a rolling
upgrade?
does the java 1.0.1 client library talk ok to a 1.1.0 cluster?
I'm really excited about some of the new stuff in 1.1.0...
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.