raboof/pekko
1
0
Fork 0
Code Issues Pull requests Projects Releases 6 Packages Wiki Activity Actions 10

Use consistent hash to heartbeat to a few nodes instead of all, see #2284

Browse source 
* Previously heartbeat messages was sent to all other members, i.e.
  each member was monitored by all other members in the cluster.
* This was the number one know scalability bottleneck, due to the
  number of interconnections.
* Limit sending of heartbeats to a few (5) members. Select and
  re-balance with consistent hashing algorithm when new members
  are added or removed.
* Send a few EndHeartbeat when ending send of Heartbeat messages.
This commit is contained in:
Patrik Nordwall 2012-10-01 14:12:20 +02:00
parent 7557433491
commit 3f73705abc
8 changed files with 172 additions and 60 deletions
Download patch file Download diff file Expand all files Collapse all files

4
akka-cluster/src/main/scala/akka/cluster/Cluster.scala
View file

@ -62,7 +62,7 @@ class Cluster(val system: ExtendedActorSystem) extends Extension {
val settings = new ClusterSettings(system.settings.config, system.name)
import settings._
val selfAddress = system.provider match {
val selfAddress: Address = system.provider match {
case c: ClusterActorRefProvider ⇒ c.transport.address
case other ⇒ throw new ConfigurationException(
"ActorSystem [%s] needs to have a 'ClusterActorRefProvider' enabled in the configuration, currently uses [%s]".
@ -74,7 +74,7 @@ class Cluster(val system: ExtendedActorSystem) extends Extension {
log.info("Cluster Node [{}] - is starting up...", selfAddress)
val failureDetector = {
val failureDetector: FailureDetector = {
import settings.{ FailureDetectorImplementationClass ⇒ fqcn }
system.dynamicAccess.createInstanceFor[FailureDetector](
fqcn, Seq(classOf[ActorSystem] -> system, classOf[ClusterSettings] -> settings)).recover({