[ 
https://issues.apache.org/jira/browse/IGNITE-27346?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18045169#comment-18045169
 ] 

Roman Puchkovskiy commented on IGNITE-27346:
--------------------------------------------

A suspected scenario is the following:
 # 3-node cluster with 3-node CMG and zone with replication factor equal to 3 
is started
 # Node A saves its LWM (L) to the Vault
 # Node A is stopped and remains stopped for some time
 # Other 2 nodes exclude A from the logical topology and from all assignments
 # Other 2 nodes compact the Catalog so that the earliest version's activation 
timestamp is now higher than L
 # Node A starts; on recovery, it gets the Catalog snapshot entry and applies 
it truncating the catalog, so its local earliest catalog version's activation 
ts is also higher than L
 # PartitionReplicaLifecycleManager starts, tries to get the Catalog version at 
ts=L and fails

> CatalogNotFoundException on PartitionReplicaLifecycleManager recovery
> ---------------------------------------------------------------------
>
>                 Key: IGNITE-27346
>                 URL: https://issues.apache.org/jira/browse/IGNITE-27346
>             Project: Ignite
>          Issue Type: Bug
>            Reporter: Roman Puchkovskiy
>            Priority: Major
>              Labels: ignite-3
>
> CatalogNotFoundException: IGN-CATALOG-1 Catalog not found for given 
> timestamp: 115645238121136128
> Caused by: java.util.concurrent.CompletionException: 
>   org.apache.ignite.internal.catalog.CatalogNotFoundException: 
>     IGN-CATALOG-1 Catalog not found for given timestamp: 115645238121136128 
> TraceId:b8c2a3eb
>         
>     at 
> org.apache.ignite.internal.catalog.CatalogManagerImpl.catalogAt(CatalogManagerImpl.java:217)
>     at 
> org.apache.ignite.internal.catalog.CatalogManagerImpl.activeCatalogVersion(CatalogManagerImpl.java:169)
>     at 
> org.apache.ignite.internal.partition.replicator.PartitionReplicaLifecycleManager.processZonesOnStart(PartitionReplicaLifecycleManager.java:443)
>     at 
> org.apache.ignite.internal.partition.replicator.PartitionReplicaLifecycleManager.startAsync(PartitionReplicaLifecycleManager.java:418)



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to