Re: Review Request 66538: Added unit test slave recovery for default executor tests.

2018-04-12 Thread Qian Zhang 

---
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/66538/#review200987
---


Fix it, then Ship it!





src/tests/default_executor_tests.cpp
Lines 1375 (patched)


We are now verifying agent recovery rather than executor re-register. So I 
think the comments should be something like:
```
// This test verifies that the agent could recover if the agent
// metadata is checkpointed
```


- Qian Zhang


On April 12, 2018, 4:40 a.m., Gilbert Song wrote:
> 
> ---
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/66538/
> ---
> 
> (Updated April 12, 2018, 4:40 a.m.)
> 
> 
> Review request for mesos, Jie Yu, Kevin Klues, Qian Zhang, and Vinod Kone.
> 
> 
> Bugs: MESOS-8772
> https://issues.apache.org/jira/browse/MESOS-8772
> 
> 
> Repository: mesos
> 
> 
> Description
> ---
> 
> Added unit test slave recovery for default executor tests.
> 
> 
> Diffs
> -
> 
>   src/tests/default_executor_tests.cpp 
> 293dd20d882447401572835bd31e197faf76861b 
> 
> 
> Diff: https://reviews.apache.org/r/66538/diff/4/
> 
> 
> Testing
> ---
> 
> make check
> 
> 
> Thanks,
> 
> Gilbert Song
> 
>

Re: Review Request 66538: Added unit test slave recovery for default executor tests.

2018-04-11 Thread Gilbert Song 

---
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/66538/
---

(Updated April 11, 2018, 1:40 p.m.)


Review request for mesos, Jie Yu, Kevin Klues, Qian Zhang, and Vinod Kone.


Bugs: MESOS-8772
https://issues.apache.org/jira/browse/MESOS-8772


Repository: mesos


Description
---

Added unit test slave recovery for default executor tests.


Diffs (updated)
-

  src/tests/default_executor_tests.cpp 293dd20d882447401572835bd31e197faf76861b 


Diff: https://reviews.apache.org/r/66538/diff/4/

Changes: https://reviews.apache.org/r/66538/diff/3-4/


Testing
---

make check


Thanks,

Gilbert Song

Re: Review Request 66538: Added unit test slave recovery for default executor tests.

2018-04-11 Thread Gilbert Song 

---
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/66538/
---

(Updated April 11, 2018, 12:01 p.m.)


Review request for mesos, Jie Yu, Kevin Klues, Qian Zhang, and Vinod Kone.


Bugs: MESOS-8772
https://issues.apache.org/jira/browse/MESOS-8772


Repository: mesos


Description
---

Added unit test slave recovery for default executor tests.


Diffs (updated)
-

  src/tests/default_executor_tests.cpp 293dd20d882447401572835bd31e197faf76861b 


Diff: https://reviews.apache.org/r/66538/diff/3/

Changes: https://reviews.apache.org/r/66538/diff/2-3/


Testing
---

make check


Thanks,

Gilbert Song

Re: Review Request 66538: Added unit test slave recovery for default executor tests.

2018-04-11 Thread Gilbert Song 

---
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/66538/
---

(Updated April 11, 2018, 11:53 a.m.)


Review request for mesos, Jie Yu, Kevin Klues, Qian Zhang, and Vinod Kone.


Bugs: MESOS-8772
https://issues.apache.org/jira/browse/MESOS-8772


Repository: mesos


Description
---

Added unit test slave recovery for default executor tests.


Diffs (updated)
-

  src/tests/default_executor_tests.cpp 293dd20d882447401572835bd31e197faf76861b 


Diff: https://reviews.apache.org/r/66538/diff/2/

Changes: https://reviews.apache.org/r/66538/diff/1-2/


Testing
---

make check


Thanks,

Gilbert Song

Re: Review Request 66538: Added unit test slave recovery for default executor tests.

2018-04-11 Thread Gilbert Song 


> On April 11, 2018, 7:14 a.m., Qian Zhang wrote:
> > src/tests/default_executor_tests.cpp
> > Lines 1427-1430 (patched)
> > 
> >
> > Why do we need to launch the executor using reserved resources? This 
> > seems not relevant to the target of this test. I think we can just launch 
> > both the executor and the task with unreserved resources.

Ah, my mistake by copying from the test above. Thanks!


> On April 11, 2018, 7:14 a.m., Qian Zhang wrote:
> > src/tests/default_executor_tests.cpp
> > Lines 1481-1485 (patched)
> > 
> >
> > So we use this as a signal that the default executor has re-registered 
> > with the agent, right? But I think this `TASK_RUNNING` status update will 
> > actually be resent by the agent (rather than the executor) after the agent 
> > re-registered with the master and even before the executor re-registered 
> > with the agent. So I think it might not be used as such signal.

Ah, good catch. we could not verified the executor reregistration due to 
MESOS-6623


- Gilbert


---
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/66538/#review200891
---


On April 10, 2018, 4:41 p.m., Gilbert Song wrote:
> 
> ---
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/66538/
> ---
> 
> (Updated April 10, 2018, 4:41 p.m.)
> 
> 
> Review request for mesos, Jie Yu, Kevin Klues, Qian Zhang, and Vinod Kone.
> 
> 
> Bugs: MESOS-8772
> https://issues.apache.org/jira/browse/MESOS-8772
> 
> 
> Repository: mesos
> 
> 
> Description
> ---
> 
> Added unit test slave recovery for default executor tests.
> 
> 
> Diffs
> -
> 
>   src/tests/default_executor_tests.cpp 
> 293dd20d882447401572835bd31e197faf76861b 
> 
> 
> Diff: https://reviews.apache.org/r/66538/diff/1/
> 
> 
> Testing
> ---
> 
> make check
> 
> 
> Thanks,
> 
> Gilbert Song
> 
>

Re: Review Request 66538: Added unit test slave recovery for default executor tests.

2018-04-11 Thread Qian Zhang 

---
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/66538/#review200891
---




src/tests/default_executor_tests.cpp
Lines 1427-1430 (patched)


Why do we need to launch the executor using reserved resources? This seems 
not relevant to the target of this test. I think we can just launch both the 
executor and the task with unreserved resources.



src/tests/default_executor_tests.cpp
Lines 1481-1485 (patched)


So we use this as a signal that the default executor has re-registered with 
the agent, right? But I think this `TASK_RUNNING` status update will actually 
be resent by the agent (rather than the executor) after the agent re-registered 
with the master and even before the executor re-registered with the agent. So I 
think it might not be used as such signal.


- Qian Zhang


On April 11, 2018, 7:41 a.m., Gilbert Song wrote:
> 
> ---
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/66538/
> ---
> 
> (Updated April 11, 2018, 7:41 a.m.)
> 
> 
> Review request for mesos, Jie Yu, Kevin Klues, Qian Zhang, and Vinod Kone.
> 
> 
> Bugs: MESOS-8772
> https://issues.apache.org/jira/browse/MESOS-8772
> 
> 
> Repository: mesos
> 
> 
> Description
> ---
> 
> Added unit test slave recovery for default executor tests.
> 
> 
> Diffs
> -
> 
>   src/tests/default_executor_tests.cpp 
> 293dd20d882447401572835bd31e197faf76861b 
> 
> 
> Diff: https://reviews.apache.org/r/66538/diff/1/
> 
> 
> Testing
> ---
> 
> make check
> 
> 
> Thanks,
> 
> Gilbert Song
> 
>

Review Request 66538: Added unit test slave recovery for default executor tests.

2018-04-10 Thread Gilbert Song 

---
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/66538/
---

Review request for mesos, Jie Yu, Kevin Klues, Qian Zhang, and Vinod Kone.


Bugs: MESOS-8772
https://issues.apache.org/jira/browse/MESOS-8772


Repository: mesos


Description
---

Added unit test slave recovery for default executor tests.


Diffs
-

  src/tests/default_executor_tests.cpp 293dd20d882447401572835bd31e197faf76861b 


Diff: https://reviews.apache.org/r/66538/diff/1/


Testing
---

make check


Thanks,

Gilbert Song