Stephen O'Donnell created HDDS-7928:
---------------------------------------
Summary: EC: Change ContainerReplicaPendingOps to store deadline
rather than scheduled time
Key: HDDS-7928
URL: https://issues.apache.org/jira/browse/HDDS-7928
Project: Apache Ozone
Issue Type: Sub-task
Components: SCM
Reporter: Stephen O'Donnell
Assignee: Stephen O'Donnell
In order to facilitate a "Move Manager" for the balancer, we need to schedule
pending ops on the datanodes with different deadlines.
For example, as replica scheduled due to replication would have a timeout of 10
minutes.
However the balancer schedules large batches of work each hour, so replicas
scheduled by the balancer probably need a timeout of 60 minutes.
To facilitate this, we need to change containerReplicaPending ops to store the
deadline rather than the scheduled time. This also fixes another issue, in that
the ContainerReplicaPendingOps "expiry thread" had its own setting for the
timeout for all pending ops, and it is not related to the replication manager
settings.
As part of this change, some APIs into RM has been changed a little to allow
the move manager to schedule replication commands via RM, so that logic is
consolidated in a single place. A later PR that adds the MoveManager will use
these APIs.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
