[jira] [Updated] (CXF-8367) Micrometer always reports outcome=UNKNOWN on success cases

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

[jira] [Updated] (CXF-8367) Micrometer always reports outcome=UNKNOWN on success cases

Colm O hEigeartaigh (Jira)

     [ https://issues.apache.org/jira/browse/CXF-8367?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Stefan Berger updated CXF-8367:
    Summary: Micrometer always reports outcome=UNKNOWN on success cases  (was: Micrometer always reports OUTCOME=UNKNOWN)

> Micrometer always reports outcome=UNKNOWN on success cases
> ----------------------------------------------------------
>                 Key: CXF-8367
>                 URL: https://issues.apache.org/jira/browse/CXF-8367
>             Project: CXF
>          Issue Type: Bug
>          Components: JAX-WS Runtime
>    Affects Versions: 3.4.1
>            Reporter: Stefan Berger
>            Priority: Major
> When I use the new MetricsFeature introduced in CXF-8213, the metrics look like this in positive cases:
> {code:java}
> cxf_server_requests_seconds_count{exception="None",faultCode="LOGICAL_RUNTIME_FAULT",method="POST",operation="operation1",outcome="UNKNOWN",status="UNKNOWN",uri="/myendpoint",} 1.0
> cxf_server_requests_seconds_sum{exception="None",faultCode="LOGICAL_RUNTIME_FAULT",method="POST",operation="operation1",outcome="UNKNOWN",status="UNKNOWN",uri="/myendpoint",} 0.221690028
> {code}
> In this case, Message.RESPONSE_CODE is null when the MetricsMessageOutInterceptor is called.
> In success cases, this Message attribute is set in the MessageSenderEndingInterceptor when AbstractHTTPDestination.getResponseCodeFromMessage() is called.
> MetricsMessageOutInterceptor is called before that interceptor, that's why Message.RESPONSE_CODE is not set.
> There are 2 solution options, as far as I understand:
>  # Move the MetricsMessageOutInterceptor after the MessageSenderEndingInterceptor (the consequences of that are unknown to me)
>  # Adjust the StandardTags.status() method to treat null values as 200
> There's also another problem of the faultCode being LOGICAL_RUNTIME_FAULT in my success cases. This is because the MAPAggregatorImpl always calls exchange.setOutFaultMessage(). The code inside of JaxwsFaultCodeProvider.getFaultCode() assumes that ex.getOutFaultMessage() != null indicates a Fault.

This message was sent by Atlassian Jira