> On May 3, 2018, 7 p.m., Greg Mann wrote:
> > src/master/master.cpp
> > Lines 7569 (patched)
> > <https://reviews.apache.org/r/66908/diff/1/?file=2015914#file2015914line7569>
> >
> > I think the following is appropriate?
> >
> > s/after a master failover/after a master or agent failover/
I think not? This paragraph describes two different scenarios (master failover or agent failover),
and the two parts are explained separately.
How about the following -- note the different punctation,
// Below we loop over all received operations and check whether
// they are known to the master; operations can be unknown to the
// master after a master failover. To handle dropped operations on
// agent failover we explicitly track the received operations and
// compare them against the operations known to the master.
- Benjamin
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/66908/#review202361
-----------------------------------------------------------
On May 4, 2018, 12:37 p.m., Benjamin Bannier wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/66908/
> -----------------------------------------------------------
>
> (Updated May 4, 2018, 12:37 p.m.)
>
>
> Review request for mesos, Gaston Kleiman and Greg Mann.
>
>
> Bugs: MESOS-8870
> https://issues.apache.org/jira/browse/MESOS-8870
>
>
> Repository: mesos
>
>
> Description
> -------
>
> When the master receives an `UpdateSlaveMessage` after agent failover
> it previously did not correctly detect dropped operations (operations
> known to the master, but unknown to the agent) and did not trigger
> reconciliation for such operations.
>
> This patch fixes the handler in the master so that such dropped
> operations are reconciled.
>
>
> Diffs
> -----
>
> src/master/master.cpp 7a2f69c1fe2508e16c2685cd3490d5b1f59d6ac4
> src/tests/master_slave_reconciliation_tests.cpp 6bb4263323bcfd191c8e3c1ccba10a240e9ddd83
>
>
> Diff: https://reviews.apache.org/r/66908/diff/2/
>
>
> Testing
> -------
>
> `make check`
>
> The test in this patch fails without the corresponding master change, but succeeds with
it applied.
>
>
> Thanks,
>
> Benjamin Bannier
>
>
|