reactivecouchbase not recovering from lost nodes

[chenbekor]

I am trying to use a multi node configuration in the bucket definition (in my case I use it in play but I assume it the same for the core library).

I tested failure by failing one of the nodes and noticed that reactivecouchbase throws exceptions (timeout etc)

Any help is appreciated. Here is the configuration I'm using (0.3-SNAPSHOT):

couchbase {
  actorctx {
    timeout=10000
    execution-context {
      fork-join-executor {
        parallelism-factor = 4.0
        parallelism-max = 40
      }
    }
  }
  driver {
    enableoperationtimeout=true # fail scala future and print a message on operations when the actorctx.timeout is reached
}  
  buckets = [{
    host=["IP1","IP2","IP3","IP4","IP5","IP6"]
    port="8091",
.......

[nielsboldt]

What do you mean by not recovering? I have experienced a problem that sounds similar, that seems to be caused by a memory leak in AtomicOperation.

When cb nodes are slow or down, a lot of AtomicRequest piles up in the heap and are never removed