> On Jan. 15, 2015, 7:07 p.m., Jie Yu wrote:
> >
> 
> Jie Yu wrote:
>     Also, can you follow BenM's comment in r28809 and create a ticket to 
> capture the TODO. I am wondering if we end up using two files to solve the 
> TODO, what the interface should be? Maybe we should call it 
> state::update(path, resources)? Any thoughts?

One more thought! Maybe we can leverage the 'rename' system call to solve the 
TODO?

In fact, the state::checkpoint functions should be quite generic:
1) if the path has a file there, write to a temp file and use 'rename' to 
atomically update the file
2) if the path does not have a file, write it directly

Also, once we introduced checkpoint for repeated protobuf messages, I don't 
think we need to have this checkpoint(path, resources) function anymore? 
Thoughts?


- Jie


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/29918/#review68286
-----------------------------------------------------------


On Jan. 15, 2015, 6:30 p.m., Michael Park wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/29918/
> -----------------------------------------------------------
> 
> (Updated Jan. 15, 2015, 6:30 p.m.)
> 
> 
> Review request for mesos, Adam B, Benjamin Hindman, Ben Mahler, Jie Yu, and 
> Vinod Kone.
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> Introduced checkpoint function for Resources.
> 
> 
> Diffs
> -----
> 
>   src/slave/state.hpp 70777cf6ab681c29ca4df601fe47903e1dbdf41f 
>   src/slave/state.cpp a36fa53099300ee03f051b0f5eaaafe9f1da68d1 
>   src/tests/slave_recovery_tests.cpp 809822e63b05a21418cd9297c927d656d6fd871d 
> 
> Diff: https://reviews.apache.org/r/29918/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Michael Park
> 
>

Reply via email to