[
https://issues.apache.org/jira/browse/IGNITE-5816?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Alexey Goncharuk updated IGNITE-5816:
-------------------------------------
Description:
In current flushOrWait() code we acquire writtenBytes before we try to flush.
If flush fails, we wait for written be equal to expWritten.
When a segment is being closed, this leads to one thread returning from
flushOrWait early because written is updated, but expWritten does not match the
last record position.
> Race in WAL segment leading to ClosedChannelException
> -----------------------------------------------------
>
> Key: IGNITE-5816
> URL: https://issues.apache.org/jira/browse/IGNITE-5816
> Project: Ignite
> Issue Type: Bug
> Components: cache
> Affects Versions: 2.1
> Reporter: Alexey Goncharuk
> Assignee: Alexey Goncharuk
> Fix For: 2.2
>
>
> In current flushOrWait() code we acquire writtenBytes before we try to flush.
> If flush fails, we wait for written be equal to expWritten.
> When a segment is being closed, this leads to one thread returning from
> flushOrWait early because written is updated, but expWritten does not match
> the last record position.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
