[
https://issues.apache.org/jira/browse/IGNITE-18502?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Denis Chudov updated IGNITE-18502:
----------------------------------
Summary: Excessive full partition scan on node start (was: Implement a
pending update storage)
> Excessive full partition scan on node start
> -------------------------------------------
>
> Key: IGNITE-18502
> URL: https://issues.apache.org/jira/browse/IGNITE-18502
> Project: Ignite
> Issue Type: Improvement
> Reporter: Vladislav Pyatkov
> Priority: Major
> Labels: ignite-3
>
> *Motivation*
> All tables have a transactional nature, even if we don't use transactions
> explicitly. That means all entry are storied as write intents (pending
> entries) before they will be committed. Those entries are required for
> transactional protocol and storied in memory collection
> (_PartitionListener#txsPendingRowIds_). The collection can be lost due to
> restart the node where the data located. Recovery of the collection happens
> through scan all entries in the partition and can require huge time.
> *Definition of Done*
> After storage will be implemented, the pending entries' collection should be
> replaced to it.
> The storage should recover itself in order to do not scan partition on start.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
