[Tarantool-patches] [PATCH v2] test: fix flaky replication/status.test.lua

[Tarantool-patches] [PATCH v2] test: fix flaky replication/status.test.lua

[Alexander V. Tikhonov]

Found issue:

 [009] --- replication/status.result	Wed May  6 09:03:18 2020
 [009] +++ replication/status.reject	Tue May 12 15:55:09 2020
 [009] @@ -307,11 +307,12 @@
 [009]  ...
 [009]  r.upstream.status == "stopped"
 [009]  ---
 [009] -- true
 [009] +- false
 [009]  ...
 [009]  r.upstream.message:match('Duplicate') ~= nil
 [009]  ---
 [009] -- true
 [009] +- error: '[string "return r.upstream.message:match(''Duplicate'') ..."]:1: attempt
 [009] +    to index field ''message'' (a nil value)'
 [009]  ...
 [009]  test_run:cmd('switch default')
 [009]  ---

To check the upstream status and it's message need to wait until an
upstream appears. This prevents an attempt to index a nil value when
one of those functions are called before a record about a peer appears
in box.info.replication. It was observed on test:
  replication/show_error_on_disconnect
after commit
  c6bea65f8ef5f6c737cf70c0127189d0ebcbc36e ('replication: recfg with 0
quorum returns immediately').

Closes #4969
---

Github: https://github.com/tarantool/tarantool/tree/avtikhon/gh-4969-replication-status
Issue: https://github.com/tarantool/tarantool/issues/4969

V1: Forgot to commit result file.

 test/replication/status.result   | 9 +--------
 test/replication/status.test.lua | 4 +---
 2 files changed, 2 insertions(+), 11 deletions(-)

diff --git a/test/replication/status.result b/test/replication/status.result
index a1e51e435..a86f48774 100644
--- a/test/replication/status.result
+++ b/test/replication/status.result
@@ -302,14 +302,7 @@ test_run:cmd('switch replica')
 ---
 - true
 ...
-r = box.info.replication[1]
----
-...
-r.upstream.status == "stopped"
----
-- true
-...
-r.upstream.message:match('Duplicate') ~= nil
+test_run:wait_upstream(1, {status = 'stopped', message_re = 'Duplicate'})
 ---
 - true
 ...
diff --git a/test/replication/status.test.lua b/test/replication/status.test.lua
index a26fec74b..090968172 100644
--- a/test/replication/status.test.lua
+++ b/test/replication/status.test.lua
@@ -110,9 +110,7 @@ box.space._schema:insert({'dup'})
 test_run:cmd('switch default')
 box.space._schema:insert({'dup'})
 test_run:cmd('switch replica')
-r = box.info.replication[1]
-r.upstream.status == "stopped"
-r.upstream.message:match('Duplicate') ~= nil
+test_run:wait_upstream(1, {status = 'stopped', message_re = 'Duplicate'})
 test_run:cmd('switch default')
 box.space._schema:delete({'dup'})
 test_run:cmd("push filter ', lsn: [0-9]+' to ', lsn: <number>'")
-- 
2.17.1

Re: [Tarantool-patches] [PATCH v2] test: fix flaky replication/status.test.lua

[Sergey Bronnikov]

Hi, Sasha

thanks for the patch!

LGTM

On 19:34 Tue 09 Jun , Alexander V. Tikhonov wrote:
> Found issue:
> 
>  [009] --- replication/status.result	Wed May  6 09:03:18 2020
>  [009] +++ replication/status.reject	Tue May 12 15:55:09 2020
>  [009] @@ -307,11 +307,12 @@
>  [009]  ...
>  [009]  r.upstream.status == "stopped"
>  [009]  ---
>  [009] -- true
>  [009] +- false
>  [009]  ...
>  [009]  r.upstream.message:match('Duplicate') ~= nil
>  [009]  ---
>  [009] -- true
>  [009] +- error: '[string "return r.upstream.message:match(''Duplicate'') ..."]:1: attempt
>  [009] +    to index field ''message'' (a nil value)'
>  [009]  ...
>  [009]  test_run:cmd('switch default')
>  [009]  ---
> 
> To check the upstream status and it's message need to wait until an
> upstream appears. This prevents an attempt to index a nil value when
> one of those functions are called before a record about a peer appears
> in box.info.replication. It was observed on test:
>   replication/show_error_on_disconnect
> after commit
>   c6bea65f8ef5f6c737cf70c0127189d0ebcbc36e ('replication: recfg with 0
> quorum returns immediately').
> 
> Closes #4969
> ---
> 
> Github: https://github.com/tarantool/tarantool/tree/avtikhon/gh-4969-replication-status
> Issue: https://github.com/tarantool/tarantool/issues/4969
> 
> V1: Forgot to commit result file.
> 
>  test/replication/status.result   | 9 +--------
>  test/replication/status.test.lua | 4 +---
>  2 files changed, 2 insertions(+), 11 deletions(-)
> 
> diff --git a/test/replication/status.result b/test/replication/status.result
> index a1e51e435..a86f48774 100644
> --- a/test/replication/status.result
> +++ b/test/replication/status.result
> @@ -302,14 +302,7 @@ test_run:cmd('switch replica')
>  ---
>  - true
>  ...
> -r = box.info.replication[1]
> ----
> -...
> -r.upstream.status == "stopped"
> ----
> -- true
> -...
> -r.upstream.message:match('Duplicate') ~= nil
> +test_run:wait_upstream(1, {status = 'stopped', message_re = 'Duplicate'})
>  ---
>  - true
>  ...
> diff --git a/test/replication/status.test.lua b/test/replication/status.test.lua
> index a26fec74b..090968172 100644
> --- a/test/replication/status.test.lua
> +++ b/test/replication/status.test.lua
> @@ -110,9 +110,7 @@ box.space._schema:insert({'dup'})
>  test_run:cmd('switch default')
>  box.space._schema:insert({'dup'})
>  test_run:cmd('switch replica')
> -r = box.info.replication[1]
> -r.upstream.status == "stopped"
> -r.upstream.message:match('Duplicate') ~= nil
> +test_run:wait_upstream(1, {status = 'stopped', message_re = 'Duplicate'})
>  test_run:cmd('switch default')
>  box.space._schema:delete({'dup'})
>  test_run:cmd("push filter ', lsn: [0-9]+' to ', lsn: <number>'")
> -- 
> 2.17.1
> 

Re: [Tarantool-patches] [PATCH v2] test: fix flaky replication/status.test.lua

[Kirill Yukhin]

Hello,

On 09 июн 19:34, Alexander V. Tikhonov wrote:
> Found issue:
> 
>  [009] --- replication/status.result	Wed May  6 09:03:18 2020
>  [009] +++ replication/status.reject	Tue May 12 15:55:09 2020
>  [009] @@ -307,11 +307,12 @@
>  [009]  ...
>  [009]  r.upstream.status == "stopped"
>  [009]  ---
>  [009] -- true
>  [009] +- false
>  [009]  ...
>  [009]  r.upstream.message:match('Duplicate') ~= nil
>  [009]  ---
>  [009] -- true
>  [009] +- error: '[string "return r.upstream.message:match(''Duplicate'') ..."]:1: attempt
>  [009] +    to index field ''message'' (a nil value)'
>  [009]  ...
>  [009]  test_run:cmd('switch default')
>  [009]  ---
> 
> To check the upstream status and it's message need to wait until an
> upstream appears. This prevents an attempt to index a nil value when
> one of those functions are called before a record about a peer appears
> in box.info.replication. It was observed on test:
>   replication/show_error_on_disconnect
> after commit
>   c6bea65f8ef5f6c737cf70c0127189d0ebcbc36e ('replication: recfg with 0
> quorum returns immediately').
> 
> Closes #4969

LGTM.
I've checked your patch into 1.10, 2.3, 2.4 and master.

--
Regards, Kirill Yukhin