[ 
https://issues.apache.org/jira/browse/YARN-8060?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16436383#comment-16436383
 ] 

Gour Saha commented on YARN-8060:
---------------------------------

[~billie.rinaldi], thank you for the patch. Patch 3 looks good. Few comments 
from my side -

Fundamental question:
Based on some of the improvements you made, do you think it would make sense to 
change readiness_check to readiness_checks and support multiple checks (all 
AND-ed or even a mix of AND/OR if it even makes sense)?
h5. service.component.Component.java

1.
{code:java}
    maxContainerFailurePerComp = YarnServiceConf.getInt(
        CONTAINER_FAILURE_THRESHOLD, DEFAULT_CONTAINER_FAILURE_THRESHOLD,
        componentSpec.getConfiguration(), scheduler.getConfig());
{code}
Should we honor CONTAINER_FAILURE_THRESHOLD specified at the system level? I 
think it is a Service/Component level config.
h5. HttpProbe.java

1.
{code:java}
  public ProbeStatus ping(ComponentInstance instance) {
    ProbeStatus status = new ProbeStatus();
    ContainerStatus containerStatus = instance.getContainerStatus();
    if (containerStatus == null || 
ServiceUtils.isEmpty(containerStatus.getIPs())
        || StringUtils.isEmpty(containerStatus.getHost())) {
      status.fail(this, new IOException("IP is not available yet"));
      return status;
    }
{code}
The component instance name is not getting logged in the fail message. Should 
we add it?
h5. Configurations.md

I don’t think we should add {{yarn.service.framework.path}} in the Service AM 
config section. It is a system level config.
h5. YarnServiceAPI.md

Based on the doc changes in {{ReadinessCheck}} that you made in 
YarnServiceAPI.md, can you reflect those changes in 
YARN-Simplified-V1-API-Layer-For-Services.yaml as well?

> Create default readiness check for service components
> -----------------------------------------------------
>
>                 Key: YARN-8060
>                 URL: https://issues.apache.org/jira/browse/YARN-8060
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: yarn-native-services
>            Reporter: Billie Rinaldi
>            Assignee: Billie Rinaldi
>            Priority: Major
>         Attachments: YARN-8060.1.patch, YARN-8060.2.patch, YARN-8060.3.patch
>
>
> It is currently possible for a component instance to have READY status before 
> the AM retrieves an IP for the container. We should make sure the IP has been 
> retrieved before marking the instance as READY.
> This default probe could also have an option to check for a DNS entry for the 
> instance's hostname if a DNS address is provided.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

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

Reply via email to