[Tarantool-patches] [PATCH v1] replication: change space sync mode in a loop

[Sergey Bronnikov]

Hi, thanks for review!

On 01.09.2020 15:25, Serge Petrenko wrote:
>> I have rewrote tests to consider to cases:
>> - positive, when we have achievable quorum, set mode to sync or async 
>> on a master
>> and then make an insert. Expected successful writes always.
>> - negative, when we change mode to sync or async with tx's in a limbo.
>>
>> So comment above is not relevant now.
>
> Hi! Thanks for the fixes.
>
> The negative case seems pointless, since the change won't happen until 
> the limbo
> gets emptied out. This is because alter{is_sync=...} is a DML 
> operation which
> replaces an entry in _space system space, so it gets stuck on a 
> non-empty limbo,
> just like any DML op.

Yes, I had doubts regarding this testcase despite I know about such 
behavior.

> I suggest you remove it altogether.

Well, negative testcase removed.


>
> Other than that LGTM.
>
>