@warkolm just took this print:
The log is full with the same message, even with the load in the cluster very low (it's 7AM, nobody is using our cluster basically, and the number of indexed documents is also very low).
About every second:
[instance-0000000024] sending transport message [MessageSerializer{Response{30185}{false}{false}{false}{class org.elasticsearch.action.get.GetResponse}}] of size [647] on [Netty4TcpChannel{localAddress=/172.17.0.10:19190, remoteAddress=/172.26.46.202:47342, profile=default}] took [6004ms] which is above the warn threshold of [5000ms]
And sometimes it appears too for all our instances:
[instance-0000000024] GC did bring memory usage down, before [4091706208], after [2293931872], allocations [45], duration [145]