> On Dec. 19, 2014, 1:42 a.m., Ben Mahler wrote:
> > src/messages/messages.proto, lines 248-251
> > <https://reviews.apache.org/r/28723/diff/1/?file=782815#file782815line248>
> >
> >     Per our discussion, we should probably call out that these are 
> > resources that must be explicitly released by the framework.

Added.


> On Dec. 19, 2014, 1:42 a.m., Ben Mahler wrote:
> > src/messages/messages.proto, line 328
> > <https://reviews.apache.org/r/28723/diff/1/?file=782815#file782815line328>
> >
> >     Let's add a comment that this is sent to the slave whenever there is an 
> > acquisition or release of a persistent resource (and we probably need to 
> > piggy-back it on (re-)registration too? Maybe a TODO in this patch?)

Done.


- Jie


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/28723/#review65577
-----------------------------------------------------------


On Dec. 19, 2014, 6:39 p.m., Jie Yu wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/28723/
> -----------------------------------------------------------
> 
> (Updated Dec. 19, 2014, 6:39 p.m.)
> 
> 
> Review request for mesos and Ben Mahler.
> 
> 
> Bugs: MESOS-2100
>     https://issues.apache.org/jira/browse/MESOS-2100
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> Added protobuf protocol for communicating persisted resources between master 
> and slave.
> 
> The reason for introducing the protocol is because the master needs to tell 
> the slave to update it's persisted resources (for either persitent disk or 
> dynamic reservation) in the case of acquiring/releasing persistent disk or 
> dynamic reservations. Also, the slave needs to tell the master what resources 
> it has persisted during registration and re-registration.
> 
> We've been thought about by sending diffs (i.e., newly acquired/released 
> persisted resources). However, due to various race conditions between master 
> and slave, we decided to instead send totals (i.e., master sends what it 
> thinks is the total persisted resources for that slave to the slave).
> 
> 
> Diffs
> -----
> 
>   src/messages/messages.proto 28e593f338a154892b1cdf398710bb44e3f9f119 
> 
> Diff: https://reviews.apache.org/r/28723/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Jie Yu
> 
>

Reply via email to