* Since DeathWatchNotification is sent over the control channel it may overtake
other messages that have been sent from the same actor before it stopped.
* It can be confusing that Terminated can't be used as an end-of-conversation marker.
* In classic Remoting we didn't have this problem because all messages were sent over
the same connection.
* don't send DeathWatchNotification when system is terminating
* when using Cluster we can rely on that the other side will publish AddressTerminated
when the member has been removed
* it's actually already a race condition that often will result in that the DeathWatchNotification
from the terminating side
* in DeathWatch.scala it will remove the watchedBy when receiving AddressTerminated, and that
may (sometimes) happen before tellWatchersWeDied
* same for Unwatch
* to avoid sending many Unwatch messages when watcher's ActorSystem is terminated
* same race exists for Unwatch as for DeathWatchNotification, if RemoteWatcher publishAddressTerminated
before the watcher is terminated
* config for the flush timeout, and possibility to disable
* change package name to akka.cluster.sbr
* reference.conf has same config paths
* akka.cluster.sbr.SplitBrainResolverProvider instead of com.lightbend.akka.sbr.SplitBrainResolverProvider
* dependency from akka-cluster to akka-coordination, for lease strategy
* move TestLease to akka-coordination and use that in SBR tests
* remove keep-referee strategy
* use keep-majority by default
* review and adjust reference documentation
Co-authored-by: Johan Andrén <johan@markatta.com>
Co-authored-by: Johannes Rudolph <johannes.rudolph@gmail.com>
Co-authored-by: Christopher Batey <christopher.batey@gmail.com>
Co-authored-by: Arnout Engelen <github@bzzt.net>
* Add scalafix plugin for jdk 9.
* Add command alias sortImports.
* Excludes some sources from SortImports.
* Update SortImports to 0.4.0
* Sort imports with `sortImports` command.
* scalafix ExplicitNonNullaryApply prepare
+ Temporarily use com.sandinh:sbt-scalafix because scalacenter/scalafix#1098
+ Add ExplicitNonNullaryApply rule to .scalafix.conf
+ Manually fix a NonNullaryApply case in DeathWatchSpec that cause
`fixall` fail because ExplicitNonNullaryApply rule incorrectly rewrite
`context unbecome` to `context unbecome()` instead of `context.unbecome()`
* scalafix ExplicitNonNullaryApply
fix by enabling only ExplicitNonNullaryApply rule in .scalafix.conf then:
```
% sbt -Dakka.build.scalaVersion=2.13.1
> fixall
```
* scalafmtAll
* Revert to ch.epfl.scala:sbt-scalafix
Co-authored-by: Bùi Việt Thành <thanhbv@sandinh.net>
* First step of rolling update for cluster serlializer to not use class manifests #13654
* Test coverage
* A bit of mima excludes makes jenkins a happy butler
* Docs mention
* Expanding the LogMarker to optionally include Map of additional
properties
* The name of the marker shows up as `tags` in Kibana
* The properties of the LogMarker are included as MDC entries, which
Logstash encoder automatically understands
* Implemented with classic eventStream logging so that it can be used
from all places without dependencies to Slf4j
* also means that it's possible to subscribe to LogEventWithMarker
to consume these events via the eventStream
* move marker definitions to RemoteLogMarker and ClusterLogMarker
* marker for dead letters
* marker for leader detained/allowed
* marker for member status changed
* markers for shard allocated and started
* test LogMarker with properties in Slf4jLoggerSpec
* doc note of the LogMarker definitions
* Problem could be recreated with DowningProviderSpec when running
with -Dakka.remote.artery.enabled=off
* The exception from the downing provider caused cluster shutdown
followed by CoordinatedShutdown and ActorSystem.finalTerminate
* The ActorSystem had still not fully initialized so it tried to
load the configured extensions and the SystemMaterializer extension
* SystemMaterializer can't create child actor
* The thrown exception isn't wrong but might be confusing
* Skip loading extensions if ActorSystem already terminated
* The reason for different behavior with Artery and classic remoting
is that Artery inits the SystemMaterializer earlier
* Use embedded media driver for StressSpec
The current setup starts a media driver for each JVM in the same
directory which is caused the cnc file to be courruped in some way
so that the client in Akka timed out waiting for it to be initialized.
This test is disabled for Artery, see
https://github.com/akka/akka/issues/21810 so all it does is startup
and then end the test right away.
Will look at enabling the test again separately.
Fixes#27030
* some cluster logging improvements
* most logger names are actually good, when using ActorLogging since
config can be setup on the package (prefix)
* override logSource when StageLogging is used
* replace system.log with more specific logger
* moved to cluster tests, in new package akka.cluster.testkit
* changed config in tests
* migration guide
* documentation clarificiations for Downing and Leaving
* update warnings in Singleton and Sharding
* Config to exclude class prefix from serialize-messages, #24273
* Adding no-serialization-verification-needed-class-prefix, with "akka." included
by default
* This is important now when we disable Java serialization by default.
There will be many complaints about Akka internal classes that are not marked with
NoSerializationVerificationNeeded, and we can't really sprinkle that everywhere.
* Not removing NoSerializationVerificationNeeded usage in Akka classes because that
may break bin compat, even though it's needed any more.
* fix DisabledJavaSerializerWarningSpec
* see scenario in added test, extracted from debug logs "New gossip published"
from MultiDcSplitBrainSpec
* it publised UnreachableDataCenter even though it had already published that,
when a new cross link is added as unreachable in the reachability table
* don't know why the implementation of diffUnreachableDataCenter/isReachable was
so complicated, but no tests are failing for the changed implementation
