Join GitHub today
GitHub is home to over 31 million developers working together to host and review code, manage projects, and build software together.Sign up
Aggregations: DateHistogram with negative 'pre_offset' or 'post_offset' value ends with "Message not fully read (response) for" #7312
We're using version 1.3.2 and Oracle JRE: 1.7.0_67. The cluster has 3 nodes (2 data nodes and 1 used only to route searches).
There is a separate (different named) cluster of 1 node using marvel to collect statistics from the first one. No other nodes in the network, and no custom/manually made java clients around.
As the title says, when doing a DateHistogramAggregation using a negative value for either pre_offset or post_offset yields the message:
" [transport.netty ] Message not fully read (response) for  handler org.elasticsearch.search.action.SearchServiceTransportAction$6@f58070b, error [false], resetting" in the logs.
Note that every other query works flawlessly. This is only reproduceable when using the arguments pre_offset and/or post_offset in a DateHistogram with a negative value and with more than 1 node in the cluster.
Facts so far:
Any ideas? In the meantime, we solved the issue by using pre_zone and post_zone instead of pre_offset and post_offset. Negative values are ok, and everything is running smoothly with those (we tried index, search, and snapshot operations). No error messages in the logs.
Thanks in advance,