[
https://issues.apache.org/jira/browse/HDDS-17589?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
rajesh updated HDDS-17589:
--------------------------
Description:
Parent task: HDDS-17586.
On an OM read trace, compare each parent span with its children. A parent of
about 100ms whose children are about 1ms each is missing the span for the work
that used the time. Add that span.
Check:
* {{InfoVolume}}, {{InfoBucket}}, {{GetKeyInfo}} and {{ListStatus}} against the
ACL check and the manager call.
* {{GetKeyInfo}} against {{OM.readKeyInfo}}, {{OM.refreshPipeline}} and
{{OM.sortDatanodes}}.
* The OM call
{{StorageContainerLocationProtocol.getContainerWithPipelineBatch}} against the
SCM server span. Time OM spends before sending that call, or after the reply
returns, needs its own span when it is a large part of {{GetKeyInfo}}.
A read that is already explained by one SCM or ACL child does not need another
span. Add a span only for the gap between the parent and the sum of its
children.
was:
Add attributes to OM spans so that traces can be filtered and failures
understood without the audit log:
* volume, bucket and key, using the same names as the audit log
* request type and client user
* result status for failed requests (\{{ozone.status}}, for example
KEY_NOT_FOUND)
* OM role when applying a transaction (\{{om.role}}: leader or follower)
* data size where it applies, such as key create and commit
Values are set only when the current span is recording, using the
\{{TracingUtil.addAttributesToCurrentSpan}} helpers.
Summary: Find OM read spans whose children do not account for the
parent time (was: Add span attributes for OM requests)
> Find OM read spans whose children do not account for the parent time
> --------------------------------------------------------------------
>
> Key: HDDS-17589
> URL: https://issues.apache.org/jira/browse/HDDS-17589
> Project: Apache Ozone
> Issue Type: Sub-task
> Reporter: rajesh
> Assignee: rajesh
> Priority: Major
>
> Parent task: HDDS-17586.
> On an OM read trace, compare each parent span with its children. A parent of
> about 100ms whose children are about 1ms each is missing the span for the
> work that used the time. Add that span.
> Check:
> * {{InfoVolume}}, {{InfoBucket}}, {{GetKeyInfo}} and {{ListStatus}} against
> the ACL check and the manager call.
> * {{GetKeyInfo}} against {{OM.readKeyInfo}}, {{OM.refreshPipeline}} and
> {{OM.sortDatanodes}}.
> * The OM call
> {{StorageContainerLocationProtocol.getContainerWithPipelineBatch}} against
> the SCM server span. Time OM spends before sending that call, or after the
> reply returns, needs its own span when it is a large part of {{GetKeyInfo}}.
> A read that is already explained by one SCM or ACL child does not need
> another span. Add a span only for the gap between the parent and the sum of
> its children.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
