[ 
https://issues.apache.org/jira/browse/FLINK-3583?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Michał Fijołek updated FLINK-3583:
----------------------------------
    Description: 
Hello.
I can see that configuration is not visible in frontend when job is running.
screenshot: 
https://issues.apache.org/jira/secure/attachment/12791527/no-configuration-when-job-running.png
When I cancel job, configuration appears but `User configuration` is still not 
visible, although server sends `user-config`.
screenshot: http://imgur.com/GNAk0ei
rest-call: http://localhost:8081/jobs/jobId/config returns
{code}
{
    "jid": "71e47c0772c7d62b81f7e3385d429cca",
    "name": "Flink Streaming Job",
    "execution-config": {
        "execution-mode": "PIPELINED",
        "restart-strategy": "Restart with fixed delay (5000 ms). #3 restart 
attempts.",
        "job-parallelism": 1,
        "object-reuse-mode": false,
        "user-config": {
            "jobmanager.web.log.path": "./data/dummyLogFile.txt",
            "local.start-webserver": "true"
        }
    }
}
{code}

So there are two problems:
_1. User cannot see configuration when job is running which is not really 
useful_
I can see that it happens because of
{code}
ExecutionConfig ec = graph.getExecutionConfig();
if (ec != null) { ... }
{code}
 in {{JobConfigHandler}}. Why {{executionConfig}} is marked as "// ------ 
Fields that are only relevant for archived execution graphs ------------" and 
can it be initialized earlier let's say in ExecutionGraph constructor?

_2. Even when configuration is visible, frontent probably parses it badly_
This should be simple to fix

I'm happy to implement both fixes if you don't mind

  was:
Hello.
I can see that configuration is not visible in frontend when job is running.
screenshot: http://imgur.com/9pwlcLz
When I cancel job, configuration appears but `User configuration` is still not 
visible, although server sends `user-config`.
screenshot: http://imgur.com/GNAk0ei
rest-call: http://localhost:8081/jobs/jobId/config returns
{code}
{
    "jid": "71e47c0772c7d62b81f7e3385d429cca",
    "name": "Flink Streaming Job",
    "execution-config": {
        "execution-mode": "PIPELINED",
        "restart-strategy": "Restart with fixed delay (5000 ms). #3 restart 
attempts.",
        "job-parallelism": 1,
        "object-reuse-mode": false,
        "user-config": {
            "jobmanager.web.log.path": "./data/dummyLogFile.txt",
            "local.start-webserver": "true"
        }
    }
}
{code}

So there are two problems:
_1. User cannot see configuration when job is running which is not really 
useful_
I can see that it happens because of
{code}
ExecutionConfig ec = graph.getExecutionConfig();
if (ec != null) { ... }
{code}
 in {{JobConfigHandler}}. Why {{executionConfig}} is marked as "// ------ 
Fields that are only relevant for archived execution graphs ------------" and 
can it be initialized earlier let's say in ExecutionGraph constructor?

_2. Even when configuration is visible, frontent probably parses it badly_
This should be simple to fix

I'm happy to implement both fixes if you don't mind


> Configuration not visible in gui when job is running
> ----------------------------------------------------
>
>                 Key: FLINK-3583
>                 URL: https://issues.apache.org/jira/browse/FLINK-3583
>             Project: Flink
>          Issue Type: Bug
>          Components: Web Client, Webfrontend
>    Affects Versions: 1.0.0
>            Reporter: Michał Fijołek
>            Priority: Minor
>             Fix For: 1.0.1
>
>         Attachments: configuration-available-when-job-ended.png, 
> no-configuration-when-job-running.png
>
>
> Hello.
> I can see that configuration is not visible in frontend when job is running.
> screenshot: 
> https://issues.apache.org/jira/secure/attachment/12791527/no-configuration-when-job-running.png
> When I cancel job, configuration appears but `User configuration` is still 
> not visible, although server sends `user-config`.
> screenshot: http://imgur.com/GNAk0ei
> rest-call: http://localhost:8081/jobs/jobId/config returns
> {code}
> {
>     "jid": "71e47c0772c7d62b81f7e3385d429cca",
>     "name": "Flink Streaming Job",
>     "execution-config": {
>         "execution-mode": "PIPELINED",
>         "restart-strategy": "Restart with fixed delay (5000 ms). #3 restart 
> attempts.",
>         "job-parallelism": 1,
>         "object-reuse-mode": false,
>         "user-config": {
>             "jobmanager.web.log.path": "./data/dummyLogFile.txt",
>             "local.start-webserver": "true"
>         }
>     }
> }
> {code}
> So there are two problems:
> _1. User cannot see configuration when job is running which is not really 
> useful_
> I can see that it happens because of
> {code}
> ExecutionConfig ec = graph.getExecutionConfig();
> if (ec != null) { ... }
> {code}
>  in {{JobConfigHandler}}. Why {{executionConfig}} is marked as "// ------ 
> Fields that are only relevant for archived execution graphs ------------" and 
> can it be initialized earlier let's say in ExecutionGraph constructor?
> _2. Even when configuration is visible, frontent probably parses it badly_
> This should be simple to fix
> I'm happy to implement both fixes if you don't mind



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to