[Tarantool-patches] [PATCH v9 1/1] applier: filter incoming synchro packets via transaction initiator

Cyrill Gorcunov gorcunov at gmail.com
Tue Jun 15 21:08:33 MSK 2021


On Tue, Jun 15, 2021 at 08:02:59PM +0300, Cyrill Gorcunov wrote:
> On Tue, Jun 15, 2021 at 05:26:06PM +0300, Serge Petrenko wrote:
> > > +box.space.sync:select{}
> > > + | ---
> > > + | - - [1]
> > > + | ...
> > 
> > You need to wait for the space creation, just like
> > you do below. Otherwise the test'll be flaky.
> > Also, please see a comment regarding wait_lsn vs wait_cond
> > below.
> 
> As being discussed due to quorum=2 and sync space we don't need
> to wait.
> 
> > > +test_run:wait_cond(function() return                    \
> > > +                   box.space.sync ~= nil and            \
> > > +                   box.space.sync:get{1} ~= nil and     \
> > > +                   box.space.sync:get{1}[1] == 1 end, 100)
> > > + | ---
> > > + | - true
> > > + | ...
> > 
> > I suggest you use wait_lsn('replica2', 'master') here
> > instead of this bulky wait_cond.
> > First of all, it takes a single line, instead of 4 lines.
> > 
> > Secondly, you forgot to test `box.space.sync.index.pk ~= nil`, meaning
> > the test will still fail occasionally, when index creation doesn't replicate
> > in time.
> 
> You mean something like below?
> ---

I force pushed an update.


More information about the Tarantool-patches mailing list