You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mesos.apache.org by Jie Yu <yu...@gmail.com> on 2014/01/24 02:01:42 UTC
Review Request 17286: Allowed log::catchup to retry to tolerate network
blips.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/17286/
-----------------------------------------------------------
Review request for mesos and Benjamin Hindman.
Repository: mesos-git
Description
-------
In case of a network blip (zookeeper network), log::catchup could get stuck even if all replicas reemerge later. This is not ideal for some users such as log::recover.
So I introduced an optional timeout for log::catchup so that the catchup operation on each position will be retried if timeout happens.
Diffs
-----
src/log/catchup.hpp 3652830
src/log/catchup.cpp 5825eae
src/tests/log_tests.cpp e493af4
Diff: https://reviews.apache.org/r/17286/diff/
Testing
-------
make check
Thanks,
Jie Yu
Re: Review Request 17286: Allowed log::catchup to retry to tolerate network
blips.
Posted by Jie Yu <yu...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/17286/
-----------------------------------------------------------
(Updated Jan. 25, 2014, 12:57 a.m.)
Review request for mesos and Benjamin Hindman.
Changes
-------
Rebased.
Repository: mesos-git
Description
-------
In case of a network blip (zookeeper network), log::catchup could get stuck even if all replicas reemerge later. This is not ideal for some users such as log::recover.
So I introduced an optional timeout for log::catchup so that the catchup operation on each position will be retried if timeout happens.
Diffs (updated)
-----
src/log/catchup.hpp 3652830
src/log/catchup.cpp 5825eae
src/tests/log_tests.cpp e493af4
Diff: https://reviews.apache.org/r/17286/diff/
Testing
-------
make check
Thanks,
Jie Yu
Re: Review Request 17286: Allowed log::catchup to retry to tolerate network
blips.
Posted by Jie Yu <yu...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/17286/
-----------------------------------------------------------
(Updated Jan. 25, 2014, 12:19 a.m.)
Review request for mesos and Benjamin Hindman.
Changes
-------
BenH's comments
Repository: mesos-git
Description
-------
In case of a network blip (zookeeper network), log::catchup could get stuck even if all replicas reemerge later. This is not ideal for some users such as log::recover.
So I introduced an optional timeout for log::catchup so that the catchup operation on each position will be retried if timeout happens.
Diffs (updated)
-----
src/log/catchup.hpp 3652830
src/log/catchup.cpp 5825eae
src/tests/log_tests.cpp e493af4
Diff: https://reviews.apache.org/r/17286/diff/
Testing
-------
make check
Thanks,
Jie Yu
Re: Review Request 17286: Allowed log::catchup to retry to tolerate network
blips.
Posted by Jie Yu <yu...@gmail.com>.
> On Jan. 24, 2014, 1:59 a.m., Benjamin Hindman wrote:
> > src/log/catchup.hpp, line 52
> > <https://reviews.apache.org/r/17286/diff/1/?file=443548#file443548line52>
> >
> > If the caller can use an arbitrary proposal number than this should be an Option rather that assume semantics of passing '0'. Can we clean this up and update the comment please?
In a separate review.
- Jie
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/17286/#review32686
-----------------------------------------------------------
On Jan. 24, 2014, 1:01 a.m., Jie Yu wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/17286/
> -----------------------------------------------------------
>
> (Updated Jan. 24, 2014, 1:01 a.m.)
>
>
> Review request for mesos and Benjamin Hindman.
>
>
> Repository: mesos-git
>
>
> Description
> -------
>
> In case of a network blip (zookeeper network), log::catchup could get stuck even if all replicas reemerge later. This is not ideal for some users such as log::recover.
>
> So I introduced an optional timeout for log::catchup so that the catchup operation on each position will be retried if timeout happens.
>
>
> Diffs
> -----
>
> src/log/catchup.hpp 3652830
> src/log/catchup.cpp 5825eae
> src/tests/log_tests.cpp e493af4
>
> Diff: https://reviews.apache.org/r/17286/diff/
>
>
> Testing
> -------
>
> make check
>
>
> Thanks,
>
> Jie Yu
>
>
Re: Review Request 17286: Allowed log::catchup to retry to tolerate network
blips.
Posted by Benjamin Hindman <be...@berkeley.edu>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/17286/#review32686
-----------------------------------------------------------
Ship it!
src/log/catchup.hpp
<https://reviews.apache.org/r/17286/#comment61628>
If the caller can use an arbitrary proposal number than this should be an Option rather that assume semantics of passing '0'. Can we clean this up and update the comment please?
src/log/catchup.hpp
<https://reviews.apache.org/r/17286/#comment61629>
What are the circumstances where someone would want to pass None? Should just just be a Duration with a default?
src/log/catchup.cpp
<https://reviews.apache.org/r/17286/#comment61630>
s/. Retry it/, retrying
- Benjamin Hindman
On Jan. 24, 2014, 1:01 a.m., Jie Yu wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/17286/
> -----------------------------------------------------------
>
> (Updated Jan. 24, 2014, 1:01 a.m.)
>
>
> Review request for mesos and Benjamin Hindman.
>
>
> Repository: mesos-git
>
>
> Description
> -------
>
> In case of a network blip (zookeeper network), log::catchup could get stuck even if all replicas reemerge later. This is not ideal for some users such as log::recover.
>
> So I introduced an optional timeout for log::catchup so that the catchup operation on each position will be retried if timeout happens.
>
>
> Diffs
> -----
>
> src/log/catchup.hpp 3652830
> src/log/catchup.cpp 5825eae
> src/tests/log_tests.cpp e493af4
>
> Diff: https://reviews.apache.org/r/17286/diff/
>
>
> Testing
> -------
>
> make check
>
>
> Thanks,
>
> Jie Yu
>
>