[
https://issues.apache.org/jira/browse/IGNITE-23087?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18030336#comment-18030336
]
Andrey Khitrin commented on IGNITE-23087:
-----------------------------------------
There was a misconfiguration in the test. With 2 nodes, we must use
HIGH_AVAILABILITY consistency mode to make sure a data is available after
shutting down a single node. When this mode is enabled in a distribution zone,
data is available to read.
Also, READ_ONLY partition state seems to be fine.
So, we could close this ticket as invalid.
> Wrong partitions status if 1 node of 2 nodes cluster is down
> ------------------------------------------------------------
>
> Key: IGNITE-23087
> URL: https://issues.apache.org/jira/browse/IGNITE-23087
> Project: Ignite
> Issue Type: Bug
> Components: general, persistence
> Affects Versions: 3.0
> Environment: The 2 nodes cluster (1 CMG node).
> Reporter: Igor
> Priority: Major
> Labels: ignite-3
>
> *Steps to reproduce:*
> # Start cluster of *2 nodes* with *one CMG* node.
> # Create zone with replication equals to amount of nodes (2).
> # Create 10 tables inside the zone.
> # Insert 100 rows in every table.
> # Await all tables*partitions*nodes local state is "HEALTHY"
> # Await all tables*partitions*nodes global state is "AVAILABLE"
> # Kill *non CMG* node with kill -9.
> # Assert physical topology contains only 1 alive node.
> # Assert logical topology contains only 1 alive node.
> # Await all tables*partitions*nodes local state is "HEALTHY"
> # Await all tables*partitions*nodes global state is "UNAVAILABLE".
> *Expected:*
> Statuses equal expected.
> *Actual:*
> Statuses "HEALTHY" and "READ_ONLY"
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
