You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@samza.apache.org by Branislav Cogic <b....@levi9.com> on 2016/08/30 07:36:43 UTC

Review Request 51516: SAMZA-702: Document the significance of all the different metrics emitted by Samza out of the box

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/51516/
-----------------------------------------------------------

Review request for samza.


Bugs: SAMZA-702
    https://issues.apache.org/jira/browse/SAMZA-702


Repository: samza


Description
-------

All the metrics documented in a metrics-table.

Few counters and timer removed because they are not used:
"send-calls" counter and "chooser-update-ns" timer from SamzaContainerMetrics
"batch-resets" counter from BootstrapingChooserMetrics


Diffs
-----

  docs/_layouts/default.html 7beb734ddeaecb7a6369f7d2a5d4e0c67655269c 
  docs/learn/documentation/versioned/container/metrics-table.html PRE-CREATION 
  docs/learn/documentation/versioned/container/metrics.md b053b792097400536ea385cb3db720f6f71da017 
  samza-core/src/main/scala/org/apache/samza/container/SamzaContainerMetrics.scala 1e7515e8e8eb5ff2f769bea3184ce49308bada9a 
  samza-core/src/main/scala/org/apache/samza/system/chooser/BootstrappingChooser.scala 1cd8e0637e2192460a9e9fe078c735444be8eb97 

Diff: https://reviews.apache.org/r/51516/diff/


Testing
-------

Site ran locally using local-site-test.sh


Thanks,

Branislav Cogic


Re: Review Request 51516: SAMZA-702: Document the significance of all the different metrics emitted by Samza out of the box

Posted by Jagadish Venkatraman <ja...@gmail.com>.

> On Sept. 1, 2016, 1:50 a.m., Navina Ramesh wrote:
> > Overall, the patch looks pretty good! The only roadblock I see is that with SAMZA-680, we introduced "ContainerProcessManagerMetrics" which would replace "SamzaAppMasterMetrics". This is kind of problematic because it seems to expose 2 copies of essentially the same metrics - under different "groups" or "class names". I am not sure why we have 2 copies of the same metric? 
> > @vjagadish : why do we have 2 copies of these metrics? What is the plan going forward? 
> > 
> > Otherwise, +1 for this patch. Thanks a lot!

We have discussed this extensively in the past and have concluded that we'll deprecate SamzaAppMasterMetrics in a subsequent release. The release notes of the upcoming release will call it out.


- Jagadish


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/51516/#review147523
-----------------------------------------------------------


On Aug. 30, 2016, 7:36 a.m., Branislav Cogic wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/51516/
> -----------------------------------------------------------
> 
> (Updated Aug. 30, 2016, 7:36 a.m.)
> 
> 
> Review request for samza.
> 
> 
> Bugs: SAMZA-702
>     https://issues.apache.org/jira/browse/SAMZA-702
> 
> 
> Repository: samza
> 
> 
> Description
> -------
> 
> All the metrics documented in a metrics-table.
> 
> Few counters and timer removed because they are not used:
> "send-calls" counter and "chooser-update-ns" timer from SamzaContainerMetrics
> "batch-resets" counter from BootstrapingChooserMetrics
> 
> 
> Diffs
> -----
> 
>   docs/_layouts/default.html 7beb734ddeaecb7a6369f7d2a5d4e0c67655269c 
>   docs/learn/documentation/versioned/container/metrics-table.html PRE-CREATION 
>   docs/learn/documentation/versioned/container/metrics.md b053b792097400536ea385cb3db720f6f71da017 
>   samza-core/src/main/scala/org/apache/samza/container/SamzaContainerMetrics.scala 1e7515e8e8eb5ff2f769bea3184ce49308bada9a 
>   samza-core/src/main/scala/org/apache/samza/system/chooser/BootstrappingChooser.scala 1cd8e0637e2192460a9e9fe078c735444be8eb97 
> 
> Diff: https://reviews.apache.org/r/51516/diff/
> 
> 
> Testing
> -------
> 
> Site ran locally using local-site-test.sh
> 
> 
> Thanks,
> 
> Branislav Cogic
> 
>


Re: Review Request 51516: SAMZA-702: Document the significance of all the different metrics emitted by Samza out of the box

Posted by Branislav Cogic <b....@levi9.com>.

> On Sept. 1, 2016, 1:50 a.m., Navina Ramesh wrote:
> > docs/learn/documentation/versioned/container/metrics-table.html, line 795
> > <https://reviews.apache.org/r/51516/diff/1/?file=1488430#file1488430line795>
> >
> >     did you mean per-system here?

No, i meant per system strea. Maybe the description was confusing so i changed it a little bit. I hope I am not misunderstanding this metric.


- Branislav


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/51516/#review147523
-----------------------------------------------------------


On Sept. 2, 2016, 9:10 a.m., Branislav Cogic wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/51516/
> -----------------------------------------------------------
> 
> (Updated Sept. 2, 2016, 9:10 a.m.)
> 
> 
> Review request for samza.
> 
> 
> Bugs: SAMZA-702
>     https://issues.apache.org/jira/browse/SAMZA-702
> 
> 
> Repository: samza
> 
> 
> Description
> -------
> 
> All the metrics documented in a metrics-table.
> 
> Few counters and timer removed because they are not used:
> - "send-calls" counter and "chooser-update-ns" timer from SamzaContainerMetrics
> - "batch-resets" counter from BootstrapingChooserMetrics
> - "producer-retries" marked as deprecated in KafkaSystemProducerMetrics
> 
> 
> Diffs
> -----
> 
>   docs/_layouts/default.html 7beb734ddeaecb7a6369f7d2a5d4e0c67655269c 
>   docs/learn/documentation/versioned/container/metrics-table.html PRE-CREATION 
>   docs/learn/documentation/versioned/container/metrics.md b053b792097400536ea385cb3db720f6f71da017 
>   samza-core/src/main/scala/org/apache/samza/container/SamzaContainerMetrics.scala 1e7515e8e8eb5ff2f769bea3184ce49308bada9a 
>   samza-core/src/main/scala/org/apache/samza/container/TaskInstanceMetrics.scala 7bedadf6597524aeca3484f1cf70ea6889452496 
>   samza-core/src/main/scala/org/apache/samza/system/chooser/BootstrappingChooser.scala 1cd8e0637e2192460a9e9fe078c735444be8eb97 
>   samza-core/src/main/scala/org/apache/samza/task/TaskInstanceCollector.scala 3b91180eb231c5010b7e5a66ba990257717ca508 
>   samza-kafka/src/main/scala/org/apache/samza/system/kafka/KafkaSystemProducerMetrics.scala d579e7bdf151ba5e2a6b3956ceb8a50c8e8026a6 
>   samza-kafka/src/test/scala/org/apache/samza/system/kafka/TestKafkaSystemProducer.scala fab998a37ad1f9c08693684ddf6354aeea43de74 
> 
> Diff: https://reviews.apache.org/r/51516/diff/
> 
> 
> Testing
> -------
> 
> Site ran locally using local-site-test.sh
> ./gradlew clean build
> ./gradlew checkstyleMain checkstyleTest
> 
> 
> Thanks,
> 
> Branislav Cogic
> 
>


Re: Review Request 51516: SAMZA-702: Document the significance of all the different metrics emitted by Samza out of the box

Posted by Navina Ramesh <nr...@linkedin.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/51516/#review147523
-----------------------------------------------------------


Fix it, then Ship it!




Overall, the patch looks pretty good! The only roadblock I see is that with SAMZA-680, we introduced "ContainerProcessManagerMetrics" which would replace "SamzaAppMasterMetrics". This is kind of problematic because it seems to expose 2 copies of essentially the same metrics - under different "groups" or "class names". I am not sure why we have 2 copies of the same metric? 
@vjagadish : why do we have 2 copies of these metrics? What is the plan going forward? 

Otherwise, +1 for this patch. Thanks a lot!


docs/learn/documentation/versioned/container/metrics-table.html (line 538)
<https://reviews.apache.org/r/51516/#comment214709>

    With SAMZA-837, I think we removed the usage of this metric and only seems to be used as a dummy check in a unit test. Can you please mark this as deprecated and remove it from the unit test class?



docs/learn/documentation/versioned/container/metrics-table.html (line 795)
<https://reviews.apache.org/r/51516/#comment214711>

    did you mean per-system here?


- Navina Ramesh


On Aug. 30, 2016, 7:36 a.m., Branislav Cogic wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/51516/
> -----------------------------------------------------------
> 
> (Updated Aug. 30, 2016, 7:36 a.m.)
> 
> 
> Review request for samza.
> 
> 
> Bugs: SAMZA-702
>     https://issues.apache.org/jira/browse/SAMZA-702
> 
> 
> Repository: samza
> 
> 
> Description
> -------
> 
> All the metrics documented in a metrics-table.
> 
> Few counters and timer removed because they are not used:
> "send-calls" counter and "chooser-update-ns" timer from SamzaContainerMetrics
> "batch-resets" counter from BootstrapingChooserMetrics
> 
> 
> Diffs
> -----
> 
>   docs/_layouts/default.html 7beb734ddeaecb7a6369f7d2a5d4e0c67655269c 
>   docs/learn/documentation/versioned/container/metrics-table.html PRE-CREATION 
>   docs/learn/documentation/versioned/container/metrics.md b053b792097400536ea385cb3db720f6f71da017 
>   samza-core/src/main/scala/org/apache/samza/container/SamzaContainerMetrics.scala 1e7515e8e8eb5ff2f769bea3184ce49308bada9a 
>   samza-core/src/main/scala/org/apache/samza/system/chooser/BootstrappingChooser.scala 1cd8e0637e2192460a9e9fe078c735444be8eb97 
> 
> Diff: https://reviews.apache.org/r/51516/diff/
> 
> 
> Testing
> -------
> 
> Site ran locally using local-site-test.sh
> 
> 
> Thanks,
> 
> Branislav Cogic
> 
>


Re: Review Request 51516: SAMZA-702: Document the significance of all the different metrics emitted by Samza out of the box

Posted by Navina Ramesh <nr...@linkedin.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/51516/#review148832
-----------------------------------------------------------


Ship it!




Ship It!

- Navina Ramesh


On Sept. 2, 2016, 9:10 a.m., Branislav Cogic wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/51516/
> -----------------------------------------------------------
> 
> (Updated Sept. 2, 2016, 9:10 a.m.)
> 
> 
> Review request for samza.
> 
> 
> Bugs: SAMZA-702
>     https://issues.apache.org/jira/browse/SAMZA-702
> 
> 
> Repository: samza
> 
> 
> Description
> -------
> 
> All the metrics documented in a metrics-table.
> 
> Few counters and timer removed because they are not used:
> - "send-calls" counter and "chooser-update-ns" timer from SamzaContainerMetrics
> - "batch-resets" counter from BootstrapingChooserMetrics
> - "producer-retries" marked as deprecated in KafkaSystemProducerMetrics
> 
> 
> Diffs
> -----
> 
>   docs/_layouts/default.html 7beb734ddeaecb7a6369f7d2a5d4e0c67655269c 
>   docs/learn/documentation/versioned/container/metrics-table.html PRE-CREATION 
>   docs/learn/documentation/versioned/container/metrics.md b053b792097400536ea385cb3db720f6f71da017 
>   samza-core/src/main/scala/org/apache/samza/container/SamzaContainerMetrics.scala 1e7515e8e8eb5ff2f769bea3184ce49308bada9a 
>   samza-core/src/main/scala/org/apache/samza/container/TaskInstanceMetrics.scala 7bedadf6597524aeca3484f1cf70ea6889452496 
>   samza-core/src/main/scala/org/apache/samza/system/chooser/BootstrappingChooser.scala 1cd8e0637e2192460a9e9fe078c735444be8eb97 
>   samza-core/src/main/scala/org/apache/samza/task/TaskInstanceCollector.scala 3b91180eb231c5010b7e5a66ba990257717ca508 
>   samza-kafka/src/main/scala/org/apache/samza/system/kafka/KafkaSystemProducerMetrics.scala d579e7bdf151ba5e2a6b3956ceb8a50c8e8026a6 
>   samza-kafka/src/test/scala/org/apache/samza/system/kafka/TestKafkaSystemProducer.scala fab998a37ad1f9c08693684ddf6354aeea43de74 
> 
> Diff: https://reviews.apache.org/r/51516/diff/
> 
> 
> Testing
> -------
> 
> Site ran locally using local-site-test.sh
> ./gradlew clean build
> ./gradlew checkstyleMain checkstyleTest
> 
> 
> Thanks,
> 
> Branislav Cogic
> 
>


Re: Review Request 51516: SAMZA-702: Document the significance of all the different metrics emitted by Samza out of the box

Posted by Branislav Cogic <b....@levi9.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/51516/
-----------------------------------------------------------

(Updated Sept. 2, 2016, 9:10 a.m.)


Review request for samza.


Bugs: SAMZA-702
    https://issues.apache.org/jira/browse/SAMZA-702


Repository: samza


Description (updated)
-------

All the metrics documented in a metrics-table.

Few counters and timer removed because they are not used:
- "send-calls" counter and "chooser-update-ns" timer from SamzaContainerMetrics
- "batch-resets" counter from BootstrapingChooserMetrics
- "producer-retries" marked as deprecated in KafkaSystemProducerMetrics


Diffs
-----

  docs/_layouts/default.html 7beb734ddeaecb7a6369f7d2a5d4e0c67655269c 
  docs/learn/documentation/versioned/container/metrics-table.html PRE-CREATION 
  docs/learn/documentation/versioned/container/metrics.md b053b792097400536ea385cb3db720f6f71da017 
  samza-core/src/main/scala/org/apache/samza/container/SamzaContainerMetrics.scala 1e7515e8e8eb5ff2f769bea3184ce49308bada9a 
  samza-core/src/main/scala/org/apache/samza/container/TaskInstanceMetrics.scala 7bedadf6597524aeca3484f1cf70ea6889452496 
  samza-core/src/main/scala/org/apache/samza/system/chooser/BootstrappingChooser.scala 1cd8e0637e2192460a9e9fe078c735444be8eb97 
  samza-core/src/main/scala/org/apache/samza/task/TaskInstanceCollector.scala 3b91180eb231c5010b7e5a66ba990257717ca508 
  samza-kafka/src/main/scala/org/apache/samza/system/kafka/KafkaSystemProducerMetrics.scala d579e7bdf151ba5e2a6b3956ceb8a50c8e8026a6 
  samza-kafka/src/test/scala/org/apache/samza/system/kafka/TestKafkaSystemProducer.scala fab998a37ad1f9c08693684ddf6354aeea43de74 

Diff: https://reviews.apache.org/r/51516/diff/


Testing (updated)
-------

Site ran locally using local-site-test.sh
./gradlew clean build
./gradlew checkstyleMain checkstyleTest


Thanks,

Branislav Cogic


Re: Review Request 51516: SAMZA-702: Document the significance of all the different metrics emitted by Samza out of the box

Posted by Branislav Cogic <b....@levi9.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/51516/
-----------------------------------------------------------

(Updated Sept. 2, 2016, 9:06 a.m.)


Review request for samza.


Bugs: SAMZA-702
    https://issues.apache.org/jira/browse/SAMZA-702


Repository: samza


Description (updated)
-------

Updated according to Navina's and Jagadish's comments


Diffs (updated)
-----

  docs/_layouts/default.html 7beb734ddeaecb7a6369f7d2a5d4e0c67655269c 
  docs/learn/documentation/versioned/container/metrics-table.html PRE-CREATION 
  docs/learn/documentation/versioned/container/metrics.md b053b792097400536ea385cb3db720f6f71da017 
  samza-core/src/main/scala/org/apache/samza/container/SamzaContainerMetrics.scala 1e7515e8e8eb5ff2f769bea3184ce49308bada9a 
  samza-core/src/main/scala/org/apache/samza/container/TaskInstanceMetrics.scala 7bedadf6597524aeca3484f1cf70ea6889452496 
  samza-core/src/main/scala/org/apache/samza/system/chooser/BootstrappingChooser.scala 1cd8e0637e2192460a9e9fe078c735444be8eb97 
  samza-core/src/main/scala/org/apache/samza/task/TaskInstanceCollector.scala 3b91180eb231c5010b7e5a66ba990257717ca508 
  samza-kafka/src/main/scala/org/apache/samza/system/kafka/KafkaSystemProducerMetrics.scala d579e7bdf151ba5e2a6b3956ceb8a50c8e8026a6 
  samza-kafka/src/test/scala/org/apache/samza/system/kafka/TestKafkaSystemProducer.scala fab998a37ad1f9c08693684ddf6354aeea43de74 

Diff: https://reviews.apache.org/r/51516/diff/


Testing
-------

Site ran locally using local-site-test.sh


Thanks,

Branislav Cogic


Re: Review Request 51516: SAMZA-702: Document the significance of all the different metrics emitted by Samza out of the box

Posted by Navina Ramesh <nr...@linkedin.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/51516/#review147506
-----------------------------------------------------------



Still going through the list. Thanks for compiling this. This is awesome work and super-useful for the Samza community!


docs/learn/documentation/versioned/container/metrics-table.html (line 231)
<https://reviews.apache.org/r/51516/#comment214696>

    Can you rephrase the description as "Number of commit method calls at TaskInstance level" so as to clearly distinguish it from Container's commit method call metrics? I understand it belongs to a different group. It will be better to distinguish them clearly.
    
    Or you can add a subtitle common to all metrics under "TaskInstanceMetrics" implying that the following apply to each TaskInstance



docs/learn/documentation/versioned/container/metrics-table.html (line 243)
<https://reviews.apache.org/r/51516/#comment214697>

    nit: "Number of messages actually processed by a task"



docs/learn/documentation/versioned/container/metrics-table.html (line 247)
<https://reviews.apache.org/r/51516/#comment214698>

    I actually don't see any difference between "send-calls" and "messages-sent". Both are technically the same. We should consider removing one of them. At least, deprecating one of them in the current version.


- Navina Ramesh


On Aug. 30, 2016, 7:36 a.m., Branislav Cogic wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/51516/
> -----------------------------------------------------------
> 
> (Updated Aug. 30, 2016, 7:36 a.m.)
> 
> 
> Review request for samza.
> 
> 
> Bugs: SAMZA-702
>     https://issues.apache.org/jira/browse/SAMZA-702
> 
> 
> Repository: samza
> 
> 
> Description
> -------
> 
> All the metrics documented in a metrics-table.
> 
> Few counters and timer removed because they are not used:
> "send-calls" counter and "chooser-update-ns" timer from SamzaContainerMetrics
> "batch-resets" counter from BootstrapingChooserMetrics
> 
> 
> Diffs
> -----
> 
>   docs/_layouts/default.html 7beb734ddeaecb7a6369f7d2a5d4e0c67655269c 
>   docs/learn/documentation/versioned/container/metrics-table.html PRE-CREATION 
>   docs/learn/documentation/versioned/container/metrics.md b053b792097400536ea385cb3db720f6f71da017 
>   samza-core/src/main/scala/org/apache/samza/container/SamzaContainerMetrics.scala 1e7515e8e8eb5ff2f769bea3184ce49308bada9a 
>   samza-core/src/main/scala/org/apache/samza/system/chooser/BootstrappingChooser.scala 1cd8e0637e2192460a9e9fe078c735444be8eb97 
> 
> Diff: https://reviews.apache.org/r/51516/diff/
> 
> 
> Testing
> -------
> 
> Site ran locally using local-site-test.sh
> 
> 
> Thanks,
> 
> Branislav Cogic
> 
>


Re: Review Request 51516: SAMZA-702: Document the significance of all the different metrics emitted by Samza out of the box

Posted by Navina Ramesh <nr...@linkedin.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/51516/#review147518
-----------------------------------------------------------




docs/learn/documentation/versioned/container/metrics-table.html (line 367)
<https://reviews.apache.org/r/51516/#comment214701>

    I think this metric just represents the number of partitions of a particular system that were empty and were provided to the consumer to poll for new messages. I don't think the correlation between $system-ssp-fetches-per-poll and $system-polls holds. Please correct me if I have misunderstood this.



docs/learn/documentation/versioned/container/metrics-table.html (line 375)
<https://reviews.apache.org/r/51516/#comment214703>

    nit: number of messages that were chosen (by the MessageChooser) for a particular system stream partition



docs/learn/documentation/versioned/container/metrics-table.html (line 379)
<https://reviews.apache.org/r/51516/#comment214705>

    nit: Average time spent polling all underlying systems for new messages (in nanoseconds)


- Navina Ramesh


On Aug. 30, 2016, 7:36 a.m., Branislav Cogic wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/51516/
> -----------------------------------------------------------
> 
> (Updated Aug. 30, 2016, 7:36 a.m.)
> 
> 
> Review request for samza.
> 
> 
> Bugs: SAMZA-702
>     https://issues.apache.org/jira/browse/SAMZA-702
> 
> 
> Repository: samza
> 
> 
> Description
> -------
> 
> All the metrics documented in a metrics-table.
> 
> Few counters and timer removed because they are not used:
> "send-calls" counter and "chooser-update-ns" timer from SamzaContainerMetrics
> "batch-resets" counter from BootstrapingChooserMetrics
> 
> 
> Diffs
> -----
> 
>   docs/_layouts/default.html 7beb734ddeaecb7a6369f7d2a5d4e0c67655269c 
>   docs/learn/documentation/versioned/container/metrics-table.html PRE-CREATION 
>   docs/learn/documentation/versioned/container/metrics.md b053b792097400536ea385cb3db720f6f71da017 
>   samza-core/src/main/scala/org/apache/samza/container/SamzaContainerMetrics.scala 1e7515e8e8eb5ff2f769bea3184ce49308bada9a 
>   samza-core/src/main/scala/org/apache/samza/system/chooser/BootstrappingChooser.scala 1cd8e0637e2192460a9e9fe078c735444be8eb97 
> 
> Diff: https://reviews.apache.org/r/51516/diff/
> 
> 
> Testing
> -------
> 
> Site ran locally using local-site-test.sh
> 
> 
> Thanks,
> 
> Branislav Cogic
> 
>