Vladimir Dmitrienko created IGNITE-25429:
--------------------------------------------
Summary: Aipersist / RocksDB: READ_ONLY partitions after 1 node
restart in 2 nodes cluster
Key: IGNITE-25429
URL: https://issues.apache.org/jira/browse/IGNITE-25429
Project: Ignite
Issue Type: Bug
Components: persistence
Affects Versions: 3.0.0-beta1
Environment: 2 nodes (one CMG, each node {color:#067d17}"-Xms512m",
"-Xmx1536m{color}{color:#067d17}"{color}), each on separate host. Each host
vCPU: 10, Memory: 16GB.
Reporter: Vladimir Dmitrienko
Attachments: cluster_logs.zip
*Steps to reproduce:*
# Create a 2 nodes cluster with 1 CMG node.
# Create zone with replication equals to amount of nodes (2).
# Create 1 table inside the zone.
# Insert 10 rows in every table.
# Await all nodes tables partitions local state is "HEALTHY"
# Await all nodes tables partitions global state is "AVAILABLE"
# Kill the non CMG node with kill -9.
# Start killed node.
# Using REST API await *physical* topology has 2 alive nodes.
# Using REST API await *logical* topology has 2 alive nodes.
# Await all nodes tables partitions local state is "HEALTHY" (by connecting to
REST of the CMG node).
# Await all nodes tables partitions local state is "AVAILABLE" (by connecting
to REST of the CMG node).
*Expected:*
All partitions become "AVAILABLE".
*Actual:*
Partitions in state: "READ_ONLY"
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
