On a few instances, I'm seeing repeated log messages in the form:
[2012-03-06 16:57:40,079][WARN ][transport.netty ]
[index2.qa.acx] Message not fully read (request) for [38951] and
action [], resetting
[2012-03-06 16:57:42,431][WARN ][transport.netty ]
[index2.qa.acx] Message not fully read (request) for [147443] and
action [], resetting
[2012-03-06 16:57:42,748][WARN ][transport.netty ]
[index2.qa.acx] Message not fully read (request) for [146541] and
action [], resetting
[2012-03-06 16:57:44,980][WARN ][transport.netty ]
[index2.qa.acx] Message not fully read (request) for [36339] and
action [], resetting
[2012-03-06 16:57:45,080][WARN ][transport.netty ]
[index2.qa.acx] Message not fully read (request) for [38953] and
action [], resetting
[2012-03-06 16:57:47,434][WARN ][transport.netty ]
[index2.qa.acx] Message not fully read (request) for [147445] and
action [], resetting
[2012-03-06 16:57:47,772][WARN ][transport.netty ]
[index2.qa.acx] Message not fully read (request) for [146543] and
action [], resetting
[2012-03-06 16:57:50,013][WARN ][transport.netty ]
[index2.qa.acx] Message not fully read (request) for [36341] and
action [], resetting
Each node in the cluster logs similar (but not exactly the same
number) messages.
I don't see this in instances that have been "nuked" while upgrading,
IE, testing instances where the data directory has been cleared while
upgrading. Only on instances that have seen an upgrade from 18.7.
O