[
https://issues.apache.org/jira/browse/HDDS-9081?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Duong updated HDDS-9081:
------------------------
Description:
Replication sends messages faster than the network/receiver can consume. This
results in many messages being queued and the amount allocated memory surges
and results in OOME.
gRPC streaming allows optimizations to minimize the queued messages by
observing the flow control.
was:Replication sends messages faster than the network/receiver can consume.
This results in many messages are queued and the amount allocated memory surges
and results in OOME. That’s aligned with what we see from assessing OOME heap
dumps (saw a lot of ByteBuffers (GB) in the Netty HTTP queue).
> Handling Netty back pressure when streaming containers for replication.
> -----------------------------------------------------------------------
>
> Key: HDDS-9081
> URL: https://issues.apache.org/jira/browse/HDDS-9081
> Project: Apache Ozone
> Issue Type: Improvement
> Reporter: Duong
> Priority: Major
>
> Replication sends messages faster than the network/receiver can consume. This
> results in many messages being queued and the amount allocated memory surges
> and results in OOME.
> gRPC streaming allows optimizations to minimize the queued messages by
> observing the flow control.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
