Re: Review Request: When sending NoMasterDetectedMessage, we need to clear the master sequence number in order to ensure we still send NewMasterDetectedMessage if the master is the same. Otherwise, we can get stuck in a master-less state.

[Ben Mahler <be...@gmail.com>]

> On March 29, 2013, 7:39 p.m., Vinod Kone wrote:
> > src/detector/detector.cpp, lines 250-252
> > <https://reviews.apache.org/r/10161/diff/1/?file=275661#file275661line250>
> >
> >     isn't this redundant?

Redundant but explicit. I'll drop them since I'm not sure what our convention is.


- Ben


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/10161/#review18529
-----------------------------------------------------------


On March 28, 2013, 6:25 p.m., Ben Mahler wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/10161/
> -----------------------------------------------------------
> 
> (Updated March 28, 2013, 6:25 p.m.)
> 
> 
> Review request for mesos, Benjamin Hindman and Vinod Kone.
> 
> 
> Description
> -------
> 
> This is also along the way to fixing MESOS-305.
> 
> This fixes a bug where we were not clearing the sequence number when sending a NoMasterDetectedMessage.
> As a result, this removes the need for hiding NoMasterDetectedMessages from contending non-leaders.
> 
> 
> Diffs
> -----
> 
>   src/detector/detector.cpp 7a8355162d543e017505dd58efd2d7bf96f99623 
> 
> Diff: https://reviews.apache.org/r/10161/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> I'd like to add a test for this, but I don't see how to induce a timedout() call on the ZooKeeperMasterDetectorProcess, unless I add some plumbing to get a handle to the underlying process. Suggestions?
> 
> 
> Thanks,
> 
> Ben Mahler
> 
>