Sergey Soldatov created HDDS-16323:
--------------------------------------
Summary: StreamBlockInputStream: refill the pre-read window in
bulk instead of once per response
Key: HDDS-16323
URL: https://issues.apache.org/jira/browse/HDDS-16323
Project: Apache Ozone
Issue Type: Bug
Components: Ozone Datanode
Affects Versions: 2.2.1
Reporter: Sergey Soldatov
Assignee: Sergey Soldatov
When streaming block reads are enabled, the client sends a new ReadBlock
request after almost every response, even though it has already requested a
large pre-read window.
With the default settings, this can mean one extra request for each 1 MiB
response during a sequential read. Each request makes the datanode repeat
request processing and block metadata lookup work, increasing CPU use and
slowing large downloads.
The client should avoid these unnecessary top-up requests while it still has
enough data already requested.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
