Ashish Kumar created HDDS-15024:
-----------------------------------
Summary: Introduce PendingContainerTracker in SCM to prevent
container over-allocation per DataNode
Key: HDDS-15024
URL: https://issues.apache.org/jira/browse/HDDS-15024
Project: Apache Ozone
Issue Type: Sub-task
Reporter: Ashish Kumar
Assignee: Ashish Kumar
Currently, SCM allocates containers to DataNodes without tracking in-flight
(pending) allocations. As a result, a DataNode may receive more container
allocations than it can effectively handle, especially under high load or
delayed container creation. This can lead to imbalance, resource pressure on
specific DNs, and suboptimal container distribution across the cluster.
*Proposed Solution*
Introduce a *PendingContainerTracker* in SCM to track container allocations
that are issued but not yet fully realized on DataNodes.
* Maintain a per-DataNode count of pending container allocations.
* Update the tracker when:
** A container is allocated (increment pending count).
** The container is successfully created/acknowledged by DN (decrement count).
** Times out (cleanup pending count).
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
