[Tarantool-patches] [PATCH 4/4] replication: add test with change space sync mode in a loop

Vladislav Shpilevoy v.shpilevoy at tarantool.org
Thu Nov 26 01:02:28 MSK 2020


Thanks for the patch!

> diff --git a/test/replication/qsync_sync_mode.result b/test/replication/qsync_sync_mode.result
> new file mode 100644
> index 000000000..66f6a70b2
> --- /dev/null
> +++ b/test/replication/qsync_sync_mode.result
> @@ -0,0 +1,114 @@
> +test_run:cmd('start server replica with wait=True, wait_load=True')
> + | ---
> + | - true
> + | ...
> +
> +-- Write data to a leader and enable/disable sync mode in a loop.
> +-- Testcase setup.
> +_ = box.schema.space.create('sync', {is_sync=true, engine=engine})
> + | ---
> + | ...
> +_ = box.space.sync:create_index('pk')
> + | ---
> + | ...
> +box.cfg{replication_synchro_quorum=2, replication_synchro_timeout=1000}
> + | ---
> + | ...
> +
> +-- Testcase body.
> +for i = 1,100 do                                                               \
> +    box.space.sync:alter{is_sync=random_boolean()}                             \

I thought about it more, and I realized that it does not make
much sense to test alter of kind true -> true, and false -> false.
Maybe better simply change it to negative on each iteration.

Also if the sync mode is true, wait_cond is supposed to be not
needed. That is the purpose of the sync. Currently your test will
pass even if I will simply remove is_sync option at all.

> +    box.space.sync:insert{i}                                                   \
> +    test_run:switch('replica')                                                 \
> +    test_run:wait_cond(function() return box.space.sync:get{i} ~= nil end)     \
> +    test_run:switch('default')                                                 \
> +    test_run:wait_cond(function() return box.space.sync:get{i} ~= nil end)     \
> +end
> + | ---
> + | ...
> +box.space.sync:count() -- 100
> + | ---
> + | - 100
> + | ...


More information about the Tarantool-patches mailing list