Heartbeat interval is growing too large


(Anatolii Botov) #1

I use DData to store members across some nodes( 2 for now). I noticed that if i add members on nodes fast, replicator starts reply slower and slower. At some point replicator has reach my timeout in 10 seconds and starts send me UpdateTimeout. In logs i see warnings “WARN [akka.remote.PhiAccrualFailureDetector] heartbeat interval is growing too large for address akka.tcp://tapi@127.0.0.1:11124: 4135 millis”.
I read only from local value, but writes need to be done on 2 nodes.
Is there a way to make process more faster?


(Anatolii Botov) #2

I think this could help https://github.com/akka/akka/issues/22171


(Patrik Nordwall) #3

That sounds strange. Have you tried with Akka 2.5.14 (I think we did some improvement of Distributed Data there).

What do you store? What data types? Do you have a way to reproduce that you can share? Minimized example app?


(Anatolii Botov) #4

Yes, i have tried akka 2.5.14.
I will try to make a sample and post it here.


(Anatolii Botov) #5

I cannot find a way to include code here. So i created an issue https://github.com/akka/akka/issues/25529