[GOBBLIN-1766] Define metric to measure lag from producing to consume…

[umustafi]

… change stream events

Dear Gobblin maintainers,

Please accept this PR. I understand that it will not be reviewed until I have checked off all the steps below!

JIRA

• My PR addresses the following Gobblin JIRA issues and references them in the PR title. For example, "[GOBBLIN-XXX] My Gobblin PR"
  • https://issues.apache.org/jira/browse/GOBBLIN-1766

Description

• Here are some details about my PR, including screenshots (if applicable):
  Modify event for change stream events to add the transaction identifier and produceTimestamp. Use the transaction identifier instead of timestamp to dedup events as "at least once" delivery may result in multiple change stream events with different timestamps for the same row update in mysql. Instead use the timestamp to emit a metric measuring the time between producing the event and consuming on our monitor.

Tests

• My PR adds the following unit tests OR does not need testing for this extremely good reason:

Commits

• My commits all reference JIRA issues in their subject lines, and I have squashed multiple commits if they address the same issue. In addition, my commits follow the guidelines from "How to write a good git commit message":
  1. Subject is separated from body by a blank line
  2. Subject is limited to 50 characters
  3. Subject does not end with a period
  4. Subject uses the imperative mood ("add", not "adding")
  5. Body wraps at 72 characters
  6. Body explains "what" and "why", not "how"

[umustafi]

@ZihanLi58 @phet can you review?

[codecov-commenter]

Codecov Report

Merging #3625 (53cad73) into master (85cd1f9) will increase coverage by 0.15%.
The diff coverage is 0.00%.

@@             Coverage Diff              @@
##             master    #3625      +/-   ##
============================================
+ Coverage     46.44%   46.60%   +0.15%     
- Complexity    10621    10667      +46     
============================================
  Files          2130     2133       +3     
  Lines         83303    83503     +200     
  Branches       9282     9281       -1     
============================================
+ Hits          38692    38914     +222     
+ Misses        41039    41030       -9     
+ Partials       3572     3559      -13     

Impacted Files	Coverage Δ
...pache/gobblin/runtime/kafka/HighLevelConsumer.java	0.00% <0.00%> (ø)
...apache/gobblin/runtime/metrics/RuntimeMetrics.java	0.00% <ø> (ø)
...ervice/monitoring/DagActionStoreChangeMonitor.java	0.00% <0.00%> (ø)
...lin/service/monitoring/SpecStoreChangeMonitor.java	0.00% <0.00%> (ø)
...dules/flowgraph/pathfinder/AbstractPathFinder.java	83.70% <0.00%> (-1.89%)	⬇️
...lin/elasticsearch/writer/FutureCallbackHolder.java	61.42% <0.00%> (-1.43%)	⬇️
.../java/org/apache/gobblin/runtime/api/FlowSpec.java	38.91% <0.00%> (-1.19%)	⬇️
...n/service/modules/template/StaticFlowTemplate.java	86.74% <0.00%> (-0.93%)	⬇️
...odules/orchestration/InMemoryUserQuotaManager.java	70.07% <0.00%> (-0.79%)	⬇️
...service/modules/dataset/HiveDatasetDescriptor.java	54.76% <0.00%> (ø)
... and 34 more

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[phet]

I also agree to have you confirm whether the txId is in fact what we want to key/cache on... otherwise leave a clarifying comment

[phet]

I clearly took a closer read this time and have a few more Qs... sorry to have not noticed them earlier

[phet]

seems very close!

do let me know if you find deeper rationale to prove uniform partition delay. intuitively I don't believe that can be assumed, but possibly there's a critical factor I'm not aware of

[phet]

generally looks good... will let you decide how to proceed on the measurement of non-uniform delay and whether anticipated to be an issue

[ZihanLi58]

+1