[
https://issues.apache.org/jira/browse/HDDS-1196?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16781114#comment-16781114
]
Bharat Viswanadham edited comment on HDDS-1196 at 3/1/19 12:27 AM:
-------------------------------------------------------------------
In ChillMode we have 4 rules:
# ContainerChillMode rule
# DatanodeChillMode rule (default to 1)
# HealthyPipeline rule
# Pipelines with atleast one datanode reported rule
Then we emit Chillmode status. Then immediately writes will be allowed in the
cluster
So, we want to give additional time to start the Replication Monitor thread.
(In order to avoid re-replication if data nodes are still being reported, as we
waited only for one datanode in the last rule.) So, this is the reason for this.
was (Author: bharatviswa):
In ChillMode we have 3 rules:
# ContainerChillMode rule
# DatanodeChillMode rule (default to 1)
# HealthyPipeline rule
# Pipelines with atleast one datanode reported rule
Then we emit Chillmode status. Then immediately writes will be allowed in the
cluster
So, we want to give additional time to start the Replication Monitor thread.
(In order to avoid re-replication if data nodes are still being reported, as we
waited only for one datanode in the last rule.) So, this is the reason for this.
> Add a ReplicationStartTimer class
> ----------------------------------
>
> Key: HDDS-1196
> URL: https://issues.apache.org/jira/browse/HDDS-1196
> Project: Hadoop Distributed Data Store
> Issue Type: Bug
> Components: SCM
> Reporter: Bharat Viswanadham
> Assignee: Bharat Viswanadham
> Priority: Major
>
> Condition for to Start Replication monitor thread.
> # Exit Chill mode
> # Time out (configurable value) default to 5 minutes. Additional time out is
> added to give some additional time for datanodes to report and make pipelines
> healthy.
> So, once we are out of chillmode, we fire ChillModeStatus, this
> ReplicationTimer Class will listen to that event, and wait for a configurable
> time, and then emit replicationEnabled.
>
> The current code, when we are out of chill mode, we set replication enabled.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
