Re: [Tarantool-patches] [PATCH 2/4] replication: test clear_synchro_queue() function

[Sergey Bronnikov <sergeyb@tarantool.org>]

Hi, thanks for review!

On 21.11.2020 17:41, Vladislav Shpilevoy wrote:
> Thanks for the patch!
>
> See 2 comments below.
>
> On 17.11.2020 17:13, sergeyb@tarantool.org wrote:
>> From: Sergey Bronnikov <sergeyb@tarantool.org>
>>
>> Part of #5055
>> Part of #4849
>> ---
>>   test/replication/qsync_basic.result   | 129 ++++++++++++++++++++++++++
>>   test/replication/qsync_basic.test.lua |  45 +++++++++
>>   2 files changed, 174 insertions(+)
>>
>> diff --git a/test/replication/qsync_basic.result b/test/replication/qsync_basic.result
>> index bd3c3cce1..a06f65ce1 100644
>> --- a/test/replication/qsync_basic.result
>> +++ b/test/replication/qsync_basic.result
>> @@ -637,6 +637,135 @@ box.space.sync:count()
>>    | - 0
>>    | ...
>>   
>> +--
>> +-- gh-4849: clear synchro queue on a master
>> +--
>> +test_run:switch('default')
>> + | ---
>> + | - true
>> + | ...
>> +box.cfg{replication_synchro_quorum = 3, replication_synchro_timeout = 1000}
>> + | ---
>> + | ...
>> +ok, err = nil
>> + | ---
>> + | ...
>> +f = fiber.create(function()							\
>> +    ok, err = pcall(box.space.sync.insert, box.space.sync, {10})		\
> 1. In Lua files we use spaces and 4 symbols indentation step. I
> know it would be easier if it would be like in C, but at this point
> it is too late to change without a reason.
>
> Here you apparently used tabs for '\' symbols.

Updated in a branch. I haven't put relevant patch here

because whitespaces are broken after copy-paste.

>
>> +end)
>> + | ---
>> + | ...
>> +f:status()
>> + | ---
>> + | - suspended
>> + | ...
>> +test_run:switch('replica')
>> + | ---
>> + | - true
>> + | ...
>> +test_run:wait_cond(function() return box.space.sync:get{10} ~= nil end)
>> + | ---
>> + | - true
>> + | ...
>> +test_run:switch('default')
>> + | ---
>> + | - true
>> + | ...
>> +box.cfg{replication_synchro_timeout = 0.1}
>> + | ---
>> + | ...
>> +box.ctl.clear_synchro_queue()
> 2. Current behaviour of the queue clearance is incorrect,
> because it should never rollback local rows, see
> https://github.com/tarantool/tarantool/issues/5435. It is
> fine to have this test, but better leave a comment, that the
> test must stop working eventually, and it will be ok to
> delete or rework it.

Updated comment:

@@ -251,11 +251,16 @@ box.space.sync:count()
  --
  -- gh-4849: clear synchro queue on a master
  --
+-- NOTE: current behavior of the queue clearance is incorrect,
+-- because it should never rollback local rows, see
+-- https://github.com/tarantool/tarantool/issues/5435. Test may
+-- stop working eventually.
+--
  test_run:switch('default')
  box.cfg{replication_synchro_quorum = 3, replication_synchro_timeout = 
1000}
  ok, err = nil