0.19.10: zip / tar.gz / issues

Release Notes:

Upgrading from 0.18 requires issuing a full flush of all the indices in the cluster (curl host:9200/_flush) before shutting down the cluster, with no indexing operations happening after the flush.
The local gateway state structure has changed from 0.18, a backup of the state files is created when upgrading, they can then be used to downgrade back to 0.18. Don’t downgrade without using them.

Upgrade to netty 3.5.8 (#2299)
Add better error handling for has_child, has_parent and top_children. (#2261)
Improve has_child filter / query performance (#2251)
Upgrade to netty 3.5.7 (#2240)
Upgrade to mvel 2.1.1 (#2228)
Disable allocation: New indices allocation not to be disabled by default (#2258)
Nodes level API: Allow to specify data:true to only execute the APIs on data nodes (#2232)
Shard Allocation: index.routing.allocation.... settings do not “remove” the setting on empty string (#2229)
Update Settings API: Allow body request to be wrapped with settings element to conform with other APIs (#2227)

has_child / has_parent filters can throw NPE when either child or parent documents aren’t indexed yet (#2297)
query_string on multiple fields with “*” fails in 0.19.9 (#2296)
Bulk Delete item when broadcast to all shard (lack of routing) might not be applied correctly (#2285)
custom_filters_score with score_mode first doesn’t return explanation correctly (#2283)
When configuring mmapfs it is not used since 0.19.7 (#2253)
failed shard while its being the source of relocation can cause inconsistent cluster state (#2242)
Field boosts not used in multi_match query (#2241)
Explain api opens 2 engine searchers, but closes only 1 engine searcher, causing resource leak (#2206)
0.19.9 no longer supports /all/status (#2205)
IndexMissingException when calling _mlt on client node (#2197)