[
https://issues.apache.org/jira/browse/KAFKA-16401?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
mutu updated KAFKA-16401:
-------------------------
Description:
The cluster consists of 3 nodes. When the storeOffsets is delay due to
fail-slow disk, the corresponding thread is blocked and holds a lock. However,
the retry requests is blocked since they cannot acquire the lock. The coming
producer and consumer requests is blocked util the fail-slow disk disappears.
Are there any comments to figure out this issues？ I will very appreciate them.
was:
The cluster consists of 3 nodes. When the storeOffsets is delay due to
fail-slow disk, the corresponding thread is blocked and holds a lock.The retry
requests consume the IO thread and network threads set in server.proterty.
However, the retry requests is blocked since they cannot acquire the lock. The
coming client requests and cosumer-pref tests failed.
Are there any comments to figure out this issues？ I will very appreciate them.
> Producer and consumer requests is blocked.
> ------------------------------------------
>
> Key: KAFKA-16401
> URL: https://issues.apache.org/jira/browse/KAFKA-16401
> Project: Kafka
> Issue Type: Bug
> Affects Versions: 2.8.2
> Reporter: mutu
> Priority: Major
> Attachments: system1.log, system2.log, system3.log
>
>
> The cluster consists of 3 nodes. When the storeOffsets is delay due to
> fail-slow disk, the corresponding thread is blocked and holds a lock.
> However, the retry requests is blocked since they cannot acquire the lock.
> The coming producer and consumer requests is blocked util the fail-slow disk
> disappears.
> Are there any comments to figure out this issues？ I will very appreciate them.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
