[jira] [Assigned] (IGNITE-16065) Cluster startup will not finish for an infinite time if starts more that 2 nodes with default partitions count.

["Evgeny Stanilovsky (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/IGNITE-16065?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Evgeny Stanilovsky reassigned IGNITE-16065:
-------------------------------------------

    Assignee: Evgeny Stanilovsky

> Cluster startup will not finish for an infinite time if starts more that 2 nodes with default partitions count.
> ---------------------------------------------------------------------------------------------------------------
>
>                 Key: IGNITE-16065
>                 URL: https://issues.apache.org/jira/browse/IGNITE-16065
>             Project: Ignite
>          Issue Type: Bug
>            Reporter: Evgeny Stanilovsky
>            Assignee: Evgeny Stanilovsky
>            Priority: Major
>              Labels: ignite-3
>
> All currently running integration tests are creates tables with limited partitions and replicas counter for example :
> ItTablesApiTest#createTable
> {noformat}
>         return node.tables().createTable(
>                 schemaName + "." + shortTableName,
>                 tblCh -> convert(SchemaBuilders.tableBuilder(schemaName, shortTableName).columns(
>                     cols).withPrimaryKey("key").build(), tblCh).changeReplicas(2).changePartitions(10)
>         );
> {noformat}
> Attempt to start more than 2 nodes and caches with default partitions counter will failed for now.
> Whole memory usage: 11,2g - 100 partitions, 24,0g - 500 partitions. Seems raft group initialization consume huge part of this mem.
> {noformat}
> org.apache.ignite.internal.table.distributed.TableManager#createTableLocally
> ...cut...
> raftMgr.prepareRaftGroup( and so on ..
> {noformat}



--
This message was sent by Atlassian Jira
(v8.20.1#820001)