You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Alexey Scherbakov (Jira)" <ji...@apache.org> on 2021/11/21 13:53:00 UTC
[jira] [Updated] (IGNITE-15965) Cluster has not been started invariant is broken
[ https://issues.apache.org/jira/browse/IGNITE-15965?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Alexey Scherbakov updated IGNITE-15965:
---------------------------------------
Description:
If [1] is called from a message handler just after the cluster start, it will trigger the assertion:
{code:java}
assert localMember != null : "Cluster has not been started";{code}
This is possible because local member is assigned after a {_}cluster.startAwait(){_}, but messages start to process a bit earlier.
Suggested fix - always delegate localMember() call to ClusterImpl.
[1] org.apache.ignite.network.scalecube.ScaleCubeTopologyService#localMember
was:
If [1] is called from a message handler just after the cluster start, it will trigger the assertion:
{code:java}
assert localMember != null : "Cluster has not been started";{code}
This is possible because local member is assigned after a _cluster.startAwait()_, but messages start to process a bit earlier.
Reproducible, for example, by [2]
Suggested fix - always delegate localMember() call to ClusterImpl.
[1] org.apache.ignite.network.scalecube.ScaleCubeTopologyService#localMember
[2] org.apache.ignite.raft.jraft.core.ItCliServiceTest#testChangePeers
> Cluster has not been started invariant is broken
> ------------------------------------------------
>
> Key: IGNITE-15965
> URL: https://issues.apache.org/jira/browse/IGNITE-15965
> Project: Ignite
> Issue Type: Bug
> Reporter: Alexey Scherbakov
> Priority: Major
> Labels: ignite-3
> Fix For: 3.0.0-alpha4
>
>
> If [1] is called from a message handler just after the cluster start, it will trigger the assertion:
> {code:java}
> assert localMember != null : "Cluster has not been started";{code}
> This is possible because local member is assigned after a {_}cluster.startAwait(){_}, but messages start to process a bit earlier.
> Suggested fix - always delegate localMember() call to ClusterImpl.
> [1] org.apache.ignite.network.scalecube.ScaleCubeTopologyService#localMember
>
--
This message was sent by Atlassian Jira
(v8.20.1#820001)