[
https://issues.apache.org/jira/browse/HADOOP-16612?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16957192#comment-16957192
]
Billie Rinaldi commented on HADOOP-16612:
-----------------------------------------
bq. Does the Hadoop's tracing system provide for sending latency info back to
the ADLS server?
Sure, the tracing information can be sent anywhere with a custom SpanReceiver
implementation, though it might be getting all the traces and not just the ABFS
ones.
bq. HTrace is EOL/unsupported. We're still working out what to do next there
Yes, I see there is a [ticket|https://jira.apache.org/jira/browse/HADOOP-15566]
open to switch to opentracing. All the tracing instrumentation libraries are
pretty similar to each other, so it won't make much of a difference.
I am okay with proceeding with this x-ms-abfs-client-latency header approach. I
wanted to make sure people are aware of the tracing possibility since it is a
good way to collect timing information, and we may want to consider
instrumenting the ABFS driver for it in the future.
> Track Azure Blob File System client-perceived latency
> -----------------------------------------------------
>
> Key: HADOOP-16612
> URL: https://issues.apache.org/jira/browse/HADOOP-16612
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/azure, hdfs-client
> Reporter: Jeetesh Mangwani
> Assignee: Jeetesh Mangwani
> Priority: Major
>
> Track the end-to-end performance of ADLS Gen 2 REST APIs by measuring latency
> in the Hadoop ABFS driver.
> The latency information is sent back to the ADLS Gen 2 REST API endpoints in
> the subsequent requests.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
