[ 
https://issues.apache.org/jira/browse/HDDS-8661?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Christos Bisias updated HDDS-8661:
----------------------------------
    Description: 
Subtask of HDDS-8538.

When a datanode dies and their containers go missing, calling 
{color:#00875a}getContainerWithPipeline() {color:#172b4d}always results in an 
exception while the pipeline isn't available. In case of dead datanodes, we 
should be using {color}getContainer() i{color:#172b4d}nstead which checks only 
the SCM and if the container doesn't exist, throws an exception. {color}{color}

When cleaning up missing containers, their datanodes aren't available and the 
containers will be only deleted from the SCM. Recon's health task should pick 
up the change and remove the container from its tables.

After trying to get a container along with a pipeline, also check just the SCM.

 

design doc: [Ozone Missing Container 
Cleanup|https://docs.google.com/document/d/1J_0D9bTCmpgqR82MYtLv9bbdrZYNB_PJdxFbgCLhUEQ/edit#]

  was:
Subtask of [HDDS-8538|https://issues.apache.org/jira/browse/HDDS-8538].

When a datanode dies and their containers go missing, calling 
{color:#00875a}getContainerWithPipeline() {color:#172b4d}always results in an 
exception while the pipeline isn't available. In case of dead datanodes, we 
should be using {color:#00875a}getContainer() {color:#172b4d}instead which 
checks only the SCM and if the container doesn't exist, throws an exception. 
{color}{color}{color}{color}

When cleaning up missing containers, their datanodes aren't available and the 
containers will be only deleted from the SCM. Recon's health task should pick 
up the change and remove the container from its tables.

After trying to get a container along with a pipeline, also check just the SCM.

 

design doc: [Ozone Missing Container 
Cleanup|https://docs.google.com/document/d/1J_0D9bTCmpgqR82MYtLv9bbdrZYNB_PJdxFbgCLhUEQ/edit#]


> Extend Recon's health task to pick up container deletes when container 
> pipeline isn't available
> -----------------------------------------------------------------------------------------------
>
>                 Key: HDDS-8661
>                 URL: https://issues.apache.org/jira/browse/HDDS-8661
>             Project: Apache Ozone
>          Issue Type: Sub-task
>            Reporter: Christos Bisias
>            Assignee: Christos Bisias
>            Priority: Major
>
> Subtask of HDDS-8538.
> When a datanode dies and their containers go missing, calling 
> {color:#00875a}getContainerWithPipeline() {color:#172b4d}always results in an 
> exception while the pipeline isn't available. In case of dead datanodes, we 
> should be using {color}getContainer() i{color:#172b4d}nstead which checks 
> only the SCM and if the container doesn't exist, throws an exception. 
> {color}{color}
> When cleaning up missing containers, their datanodes aren't available and the 
> containers will be only deleted from the SCM. Recon's health task should pick 
> up the change and remove the container from its tables.
> After trying to get a container along with a pipeline, also check just the 
> SCM.
>  
> design doc: [Ozone Missing Container 
> Cleanup|https://docs.google.com/document/d/1J_0D9bTCmpgqR82MYtLv9bbdrZYNB_PJdxFbgCLhUEQ/edit#]



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

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

Reply via email to