[jira] [Commented] (HDDS-12579) Container Replica Historical Timeline

["Sreeja (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HDDS-12579?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17946426#comment-17946426 ] 

Sreeja commented on HDDS-12579:
-------------------------------

The following are the major commands for this tool:

1. Parse Command:
{code:java}
ozone debug log container  --db=<path to db>  parse  --path=<log_folder> --thread-count=<n>{code}
Parses multiple dn-container logs concurrently.

Extracts key info: datanode ID, container ID, state, BCSID, timestamp, log level, index value, and messages (if any).

Populates two tables: DatanodeContainerLogTable(acts like a detail log history) and ContainerLogTable(stores latest state and BCSID for each container-datanode pair)

 

2. Command to show details of a single container along with analysis :
{code:java}
ozone debug log container --db=<path to db> info <container ID>{code}
Shows the complete state transition history for each replica of the given container across all datanodes where it exists.

checks for issues (only for the given container) such as:
 * Duplicate OPEN states

 * Mismatched replication

 * Under-replication or over-replication

 * Unhealthy replicas

 * Open-unhealthy replicas

 * Quasi-closed stuck containers

 

3. Command to list by final state:
{code:java}
ozone debug log container --db=<path to db> list --state=<state> [--length=<n>] [–all]{code}
Lists containers who's final state matches the given state. (only for life cycle states)

provides details such as datanode ID, container ID, BCSID, timestamp, index value, and any associated messages.

 

4. Command to list Duplicate Open Containers:
{code:java}
ozone debug log container --db=<path to db> list --duplicate-open=true{code}
identifies containers opened multiple times and provides their container ID.

 

5. Command to list OVER_REPLICATED/UNDER_REPLICATED containers:
{code:java}
ozone debug log container --db=<path to db> list-health --health-state=<type> {code}
lists containers which are OVER_REPLICATED/UNDER_REPLICATED and provides details such as datanode ID, container ID, BCSID, state, timestamp, index value, and any associated messages.

> Container Replica Historical Timeline
> -------------------------------------
>
>                 Key: HDDS-12579
>                 URL: https://issues.apache.org/jira/browse/HDDS-12579
>             Project: Apache Ozone
>          Issue Type: Epic
>            Reporter: Sreeja
>            Assignee: Sreeja
>            Priority: Major
>
> Containers are central to Ozone operations and go through various states in their lifecycle. In the past, debugging container state issues has been time-consuming due to manual steps. To optimize debugging, the goal is to provide a historical timeline and detailed container information, making it easier to identify and troubleshoot problems.
> The datanodes already log the container operations in  dn-container.log , the tool will parse these log files and store the data in a temporary DB for efficient querying.
> Hence an offline tool that would help to identify the issues related to containers by analysing the dn-container logs



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@ozone.apache.org
For additional commands, e-mail: issues-help@ozone.apache.org