[
https://issues.apache.org/jira/browse/HDDS-10516?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sammi Chen updated HDDS-10516:
------------------------------
Description:
Currently, there are following metrics on every datanode for chunk reads.
{code:java}
"numReadChunk" : 136237,
"bytesReadChunk" : 1395066880,
"LatencyReadChunkNumOps" : 136237,
"LatencyReadChunkAvgTime" : 1.0, {code}
The chunk read requests come from both remote user client and internal ratis if
the datanode is the leader of the ratis group. Currently, we cannot tell how
much data is read by user client, and how much data is read by ratis leader
itself. So this task is to introduce a new set of following metrics for read by
ratis,
{code:java}
numReadStateMachine // the request counts from ratis
bytesReadStateMachine // the total bytes read by ratis{code}
So the final goal is we want to know how much data is read by user client. So
for read from user client,
bytesReadChunk - bytesReadStateMachine = bytes read by user clients
numReadChunk - numReadStateMachine = request counts from user clients
was:
Currently, there are following metrics on every datanode for chunk reads.
{code:java}
"numReadChunk" : 136237,
"bytesReadChunk" : 1395066880,
"LatencyReadChunkNumOps" : 136237,
"LatencyReadChunkAvgTime" : 1.0, {code}
The chunk read requests come from both remote user client and internal ratis if
the datanode is the leader of the ratis group. Currently, we cannot tell how
much data is read by user client, and how much data is read by ratis leader
itself. So this task is to introduce a new set of following metrics for read by
ratis,
{code:java}
numReadStateMachine // the request counts from ratis
bytesReadStateMachine // the total bytes read by ratis{code}
For the info of read from user client,
bytesReadChunk - bytesReadStateMachine = bytes read by user clients
numReadChunk - numReadStateMachine = request counts from user clients
> Add metrics for chunk read from internal ratis
> ----------------------------------------------
>
> Key: HDDS-10516
> URL: https://issues.apache.org/jira/browse/HDDS-10516
> Project: Apache Ozone
> Issue Type: Improvement
> Reporter: Sammi Chen
> Assignee: Sammi Chen
> Priority: Major
>
> Currently, there are following metrics on every datanode for chunk reads.
> {code:java}
> "numReadChunk" : 136237,
> "bytesReadChunk" : 1395066880,
> "LatencyReadChunkNumOps" : 136237,
> "LatencyReadChunkAvgTime" : 1.0, {code}
> The chunk read requests come from both remote user client and internal ratis
> if the datanode is the leader of the ratis group. Currently, we cannot tell
> how much data is read by user client, and how much data is read by ratis
> leader itself. So this task is to introduce a new set of following metrics
> for read by ratis,
> {code:java}
> numReadStateMachine // the request counts from ratis
> bytesReadStateMachine // the total bytes read by ratis{code}
> So the final goal is we want to know how much data is read by user client. So
> for read from user client,
> bytesReadChunk - bytesReadStateMachine = bytes read by user clients
> numReadChunk - numReadStateMachine = request counts from user clients
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
