[ 
https://issues.apache.org/jira/browse/HDDS-17594?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

rajesh updated HDDS-17594:
--------------------------
    Description: 
A datanode trace today is the container-command RPC span. A key put shows 
WriteChunk and PutBlock, but not which container, which Ratis write stage, or 
why a command failed.

Close that gap:
* Chunk and block I/O, token verification and volume selection show up under 
the command span.
* Each command records {{dn.cmd_type}}, the container id, the result status and 
the exception on failure.
* A Ratis WriteChunk records its stage ({{WRITE_DATA}}, {{COMMIT_DATA}}, and 
the other stages). A write that fails because of container state records that 
state.

Depends on HDDS-17581. Reuses the same tracing helpers as HDDS-17586. No 
separate proxy, attribute or local-span sub-tasks.

The duplicate WriteChunk span from the gRPC server interceptor beside the 
dispatcher span is a separate follow-up.

  was:
A datanode trace today is the container-command RPC span. A key put shows 
WriteChunk and PutBlock, but not which container, which Ratis write stage, or 
why a command failed.

Close that gap:
* Chunk and block I/O, token verification and volume selection show up under 
the command span.
* Each command records dn.cmd\_type, the container id, the result status and 
the exception on failure.
* A Ratis WriteChunk records its stage (WRITE\_DATA, COMMIT\_DATA, and the 
other stages). A write that fails because of container state records that state.

Depends on HDDS-17581. Reuses the same tracing helpers as HDDS-17586. No 
separate proxy, attribute or local-span sub-tasks.

The duplicate WriteChunk span from the gRPC server interceptor beside the 
dispatcher span is a separate follow-up.


> Tracing improvements for Datanode
> ---------------------------------
>
>                 Key: HDDS-17594
>                 URL: https://issues.apache.org/jira/browse/HDDS-17594
>             Project: Apache Ozone
>          Issue Type: Task
>            Reporter: rajesh
>            Assignee: rajesh
>            Priority: Major
>
> A datanode trace today is the container-command RPC span. A key put shows 
> WriteChunk and PutBlock, but not which container, which Ratis write stage, or 
> why a command failed.
> Close that gap:
> * Chunk and block I/O, token verification and volume selection show up under 
> the command span.
> * Each command records {{dn.cmd_type}}, the container id, the result status 
> and the exception on failure.
> * A Ratis WriteChunk records its stage ({{WRITE_DATA}}, {{COMMIT_DATA}}, and 
> the other stages). A write that fails because of container state records that 
> state.
> Depends on HDDS-17581. Reuses the same tracing helpers as HDDS-17586. No 
> separate proxy, attribute or local-span sub-tasks.
> The duplicate WriteChunk span from the gRPC server interceptor beside the 
> dispatcher span is a separate follow-up.



--
This message was sent by Atlassian Jira
(v10.3.26#10030026)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to