[
https://issues.apache.org/jira/browse/GOBBLIN-2079?focusedWorklogId=927866&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-927866
]
ASF GitHub Bot logged work on GOBBLIN-2079:
-------------------------------------------
Author: ASF GitHub Bot
Created on: 29/Jul/24 23:34
Start Date: 29/Jul/24 23:34
Worklog Time Spent: 10m
Work Description: Will-Lo commented on code in PR #3962:
URL: https://github.com/apache/gobblin/pull/3962#discussion_r1696093690
##########
gobblin-service/src/main/java/org/apache/gobblin/service/monitoring/GaaSJobObservabilityEventProducer.java:
##########
@@ -123,7 +129,11 @@ private void setupMetrics(State state) {
public void emitObservabilityEvent(final State jobState) {
GaaSJobObservabilityEvent event = createGaaSObservabilityEvent(jobState);
- sendUnderlyingEvent(event);
+ if
(jobState.getProp(TimingEvent.FlowEventConstants.JOB_NAME_FIELD).equals(JobStatusRetriever.NA_KEY)
&& this.emitFlowObservabilityEvent) {
Review Comment:
At this point JOB_NAME_FIELD gets populated by the jobstatus monitor before
it emits the observability event
```
static Pair<org.apache.gobblin.configuration.State, NewState>
recalcJobStatus(org.apache.gobblin.configuration.State jobStatus,
StateStore<org.apache.gobblin.configuration.State> stateStore) throws
IOException {
try {
if
(!jobStatus.contains(TimingEvent.FlowEventConstants.JOB_NAME_FIELD)) {
jobStatus.setProp(TimingEvent.FlowEventConstants.JOB_NAME_FIELD,
JobStatusRetriever.NA_KEY);
}
```
and this is called right after parsing the job status
Issue Time Tracking
-------------------
Worklog Id: (was: 927866)
Time Spent: 2h 50m (was: 2h 40m)
> GaaSObservabilityEvents should have a dedicated flow level event
> ----------------------------------------------------------------
>
> Key: GOBBLIN-2079
> URL: https://issues.apache.org/jira/browse/GOBBLIN-2079
> Project: Apache Gobblin
> Issue Type: Improvement
> Reporter: William Lo
> Priority: Major
> Time Spent: 2h 50m
> Remaining Estimate: 0h
>
> GaaSJobObservabilityEvents currently encapsulate both jobs and flows in GaaS.
> This is fine for single hop flows, but flows with multiple jobs encapsulated
> in them now have a mix of job level events with the majority of metadata, and
> flow level events which provide a better view to users when their flow fails
> at any given point.
> Since the data in both events differs vastly with most metadata only having
> contextual sense in the job level event, we should separate job and flow
> level events to their own respective event types.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
