* [Tarantool-patches] [PATCH test-run 1/1] Wait_upstream/downstream should be ready to nil info
@ 2019-10-13 15:09 Vladislav Shpilevoy
2019-10-15 14:10 ` Alexander Turenko
0 siblings, 1 reply; 4+ messages in thread
From: Vladislav Shpilevoy @ 2019-10-13 15:09 UTC (permalink / raw)
To: tarantool-patches; +Cc: tarantool-patches, avtikhon, alexander.turenko
Wait_upstream/downstream are helper functions of a
test-run object. They are used to wait until a certain
state of upstream/downstream is reached such as error
message, replication status.
They take these attributes from
box.info.replication[id].upstream/downstream, which
sometimes can be nil, and it led to an error.
Now these waiters check upstream/downstream for nil.
The bug was discovered in Tarantool in
replication/show_error_on_disconnect test:
https://github.com/tarantool/tarantool/issues/4563.
It appeared, because after my patch reconfiguration of
replication with 0 quorum always returns immediately.
And right after the reconfiguration the
upstream/downstream may still do not exist. This is
why test_run:wait_upstream/downstream might fail like
this:
[002] test_run:wait_upstream(other_id, {status = 'stopped', message_re = 'Missing'})
[002] ---
[002] -- true
[002] +- error: '.../Workspaces/gitlab/test/var/002_replication/test_run.lua:250: attempt
[002] + to index local ''info'' (a nil value)'
---
Branch: https://github.com/tarantool/test-run/tree/gerold103/gh-tarantool-4563-box.info.repl.upstream-nil
Issue: https://github.com/tarantool/tarantool/issues/4563
test_run.lua | 15 ++++++++++-----
1 file changed, 10 insertions(+), 5 deletions(-)
diff --git a/test_run.lua b/test_run.lua
index 8f13491..09d53af 100644
--- a/test_run.lua
+++ b/test_run.lua
@@ -234,10 +234,15 @@ local function log_box_info_replication_cond(id, field, ok, info, opts)
status = opts.status,
message_re = opts.message_re,
})
- local got = json.encode({
- status = info.status,
- message = info.message,
- })
+ local got
+ if info then
+ got = json.encode({
+ status = info.status,
+ message = info.message,
+ })
+ else
+ got = json.encode('info is nil')
+ end
log.info('wait_%s(%d, ...); exp = %s; got = %s; result = %s', field, id,
exp, got, tostring(ok))
end
@@ -247,7 +252,7 @@ local function gen_box_info_replication_cond(id, field, opts)
local info = box.info.replication[id][field]
local ok = true
if opts.status ~= nil then
- ok = ok and info.status == opts.status
+ ok = ok and info and info.status == opts.status
end
if opts.message_re ~= nil then
-- regex match
--
2.21.0 (Apple Git-122)
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [Tarantool-patches] [PATCH test-run 1/1] Wait_upstream/downstream should be ready to nil info
2019-10-13 15:09 [Tarantool-patches] [PATCH test-run 1/1] Wait_upstream/downstream should be ready to nil info Vladislav Shpilevoy
@ 2019-10-15 14:10 ` Alexander Turenko
2019-10-15 19:09 ` [Tarantool-patches] [tarantool-patches] " Vladislav Shpilevoy
0 siblings, 1 reply; 4+ messages in thread
From: Alexander Turenko @ 2019-10-15 14:10 UTC (permalink / raw)
To: Vladislav Shpilevoy; +Cc: tarantool-patches, avtikhon, tarantool-patches
Aren't you mind against those fixes (they are mostly just my
preferences):
| diff --git a/test_run.lua b/test_run.lua
| index 09d53af..63dfdef 100644
| --- a/test_run.lua
| +++ b/test_run.lua
| @@ -234,15 +234,10 @@ local function log_box_info_replication_cond(id, field, ok, info, opts)
| status = opts.status,
| message_re = opts.message_re,
| })
| - local got
| - if info then
| - got = json.encode({
| - status = info.status,
| - message = info.message,
| - })
| - else
| - got = json.encode('info is nil')
| - end
| + local got = json.encode(info ~= nil and {
| + status = info.status,
| + message = info.message,
| + } or nil)
| log.info('wait_%s(%d, ...); exp = %s; got = %s; result = %s', field, id,
| exp, got, tostring(ok))
| end
| @@ -250,9 +245,9 @@ end
| local function gen_box_info_replication_cond(id, field, opts)
| return function()
| local info = box.info.replication[id][field]
| - local ok = true
| + local ok = info ~= nil
| if opts.status ~= nil then
| - ok = ok and info and info.status == opts.status
| + ok = ok and info.status == opts.status
| end
| if opts.message_re ~= nil then
| -- regex match
| @@ -270,6 +265,10 @@ end
|
| --- Wait for upstream status.
| ---
| +--- The function waits until information about an upstream with
| +--- provided id will appear (regardless of passed options) and
| +--- then waits for a certain state of the upstream if requested.
| +---
| --- If `opts.status` is `nil` or `box.NULL` an upstream status is
| --- not checked.
| ---
I pushed them on top of your branch as a separate commit. Squash if you are
agree.
WBR, Alexander Turenko.
On Sun, Oct 13, 2019 at 05:09:28PM +0200, Vladislav Shpilevoy wrote:
> Wait_upstream/downstream are helper functions of a
> test-run object. They are used to wait until a certain
> state of upstream/downstream is reached such as error
> message, replication status.
>
> They take these attributes from
> box.info.replication[id].upstream/downstream, which
> sometimes can be nil, and it led to an error.
>
> Now these waiters check upstream/downstream for nil.
>
> The bug was discovered in Tarantool in
> replication/show_error_on_disconnect test:
> https://github.com/tarantool/tarantool/issues/4563.
>
> It appeared, because after my patch reconfiguration of
> replication with 0 quorum always returns immediately.
> And right after the reconfiguration the
> upstream/downstream may still do not exist. This is
> why test_run:wait_upstream/downstream might fail like
> this:
>
> [002] test_run:wait_upstream(other_id, {status = 'stopped', message_re = 'Missing'})
> [002] ---
> [002] -- true
> [002] +- error: '.../Workspaces/gitlab/test/var/002_replication/test_run.lua:250: attempt
> [002] + to index local ''info'' (a nil value)'
> ---
> Branch: https://github.com/tarantool/test-run/tree/gerold103/gh-tarantool-4563-box.info.repl.upstream-nil
> Issue: https://github.com/tarantool/tarantool/issues/4563
>
> test_run.lua | 15 ++++++++++-----
> 1 file changed, 10 insertions(+), 5 deletions(-)
>
> diff --git a/test_run.lua b/test_run.lua
> index 8f13491..09d53af 100644
> --- a/test_run.lua
> +++ b/test_run.lua
> @@ -234,10 +234,15 @@ local function log_box_info_replication_cond(id, field, ok, info, opts)
> status = opts.status,
> message_re = opts.message_re,
> })
> - local got = json.encode({
> - status = info.status,
> - message = info.message,
> - })
> + local got
> + if info then
> + got = json.encode({
> + status = info.status,
> + message = info.message,
> + })
> + else
> + got = json.encode('info is nil')
> + end
> log.info('wait_%s(%d, ...); exp = %s; got = %s; result = %s', field, id,
> exp, got, tostring(ok))
> end
> @@ -247,7 +252,7 @@ local function gen_box_info_replication_cond(id, field, opts)
> local info = box.info.replication[id][field]
> local ok = true
> if opts.status ~= nil then
> - ok = ok and info.status == opts.status
> + ok = ok and info and info.status == opts.status
> end
> if opts.message_re ~= nil then
> -- regex match
> --
> 2.21.0 (Apple Git-122)
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [Tarantool-patches] [tarantool-patches] Re: [PATCH test-run 1/1] Wait_upstream/downstream should be ready to nil info
2019-10-15 14:10 ` Alexander Turenko
@ 2019-10-15 19:09 ` Vladislav Shpilevoy
2019-10-15 23:23 ` Alexander Turenko
0 siblings, 1 reply; 4+ messages in thread
From: Vladislav Shpilevoy @ 2019-10-15 19:09 UTC (permalink / raw)
To: tarantool-patches, Alexander Turenko; +Cc: avtikhon, tarantool-patches
Hi! Thanks for the review!
The changes are ok, I squashed them.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [Tarantool-patches] [tarantool-patches] Re: [PATCH test-run 1/1] Wait_upstream/downstream should be ready to nil info
2019-10-15 19:09 ` [Tarantool-patches] [tarantool-patches] " Vladislav Shpilevoy
@ 2019-10-15 23:23 ` Alexander Turenko
0 siblings, 0 replies; 4+ messages in thread
From: Alexander Turenko @ 2019-10-15 23:23 UTC (permalink / raw)
To: Vladislav Shpilevoy; +Cc: tarantool-patches, avtikhon, tarantool-patches
On Tue, Oct 15, 2019 at 09:09:50PM +0200, Vladislav Shpilevoy wrote:
> Hi! Thanks for the review!
>
> The changes are ok, I squashed them.
LGTM.
Pushed to test-run's master, updated the submodule in master, 2.2, 2.1,
1.10.
WBR, Alexander Turenko.
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2019-10-15 23:24 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-10-13 15:09 [Tarantool-patches] [PATCH test-run 1/1] Wait_upstream/downstream should be ready to nil info Vladislav Shpilevoy
2019-10-15 14:10 ` Alexander Turenko
2019-10-15 19:09 ` [Tarantool-patches] [tarantool-patches] " Vladislav Shpilevoy
2019-10-15 23:23 ` Alexander Turenko
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox