> On Nov. 18, 2014, 9:42 p.m., Adam B wrote:
> > include/mesos/mesos.proto, line 402
> > <https://reviews.apache.org/r/27550/diff/3/?file=767713#file767713line402>
> >
> >     Unique per role: Does this mean that a multi-role framework cannot 
> > change which role its persistent volume is associated with?
> >     
> >     Unique per slave: Wouldn't it be nice if you could (e.g. during a 
> > maintenance period) migrate a volume to another slave and use it there?

Unique per role: The ID is tied to the role of the resource, not the framework. 
We can change the role of a resource once we have dynamic reservation.

Unique per slave: Yes, technially it's not unique per slave, but unique per 
slave's main disk (for checkpoinging):) But to avoid confusion, I use the words 
here.


> On Nov. 18, 2014, 9:42 p.m., Adam B wrote:
> > include/mesos/mesos.proto, line 408
> > <https://reviews.apache.org/r/27550/diff/3/?file=767713#file767713line408>
> >
> >     What if host_path is set? Is it ignored (for now)? Or does it actually 
> > mean something?

It will be ignored for now. IMO, there's no need to expose 'host_path'.


> On Nov. 18, 2014, 9:42 p.m., Adam B wrote:
> > include/mesos/mesos.proto, line 412
> > <https://reviews.apache.org/r/27550/diff/3/?file=767713#file767713line412>
> >
> >     Does this necessarily describe a 'disk'? Or should it more generically 
> > be called "storage" in the future?

My opinion here is that: we still access these resources through regular posix 
vfs primitives (e.g., open/read/write), I would rather still call it disk.

If, in the future, we have some resources that provide say like key-value, blob 
store, we can call them "storage".


- Jie


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/27550/#review62016
-----------------------------------------------------------


On Nov. 18, 2014, 8:05 p.m., Jie Yu wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/27550/
> -----------------------------------------------------------
> 
> (Updated Nov. 18, 2014, 8:05 p.m.)
> 
> 
> Review request for mesos, Adam B, Benjamin Hindman, Ben Mahler, switched to 
> 'mcypark', and Vinod Kone.
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> See summary.
> 
> 
> Diffs
> -----
> 
>   include/mesos/mesos.proto 35b91180f4e317c47dfdd18e9f7fca28143ae8f5 
> 
> Diff: https://reviews.apache.org/r/27550/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Jie Yu
> 
>

Reply via email to