Tarantool development patches archive
 help / color / mirror / Atom feed
* [Tarantool-patches] [PATCH] fio: fix race condition in mktree
@ 2019-12-17 15:52 HustonMmmavr
  2020-01-10 11:33 ` Igor Munkin
  0 siblings, 1 reply; 2+ messages in thread
From: HustonMmmavr @ 2019-12-17 15:52 UTC (permalink / raw)
  To: tarantool-patches, imun

This patch fixes a bug that occours when two instances calls
fio.mktree for the same directory in parallel.

Closes #4660
---
branch: https://github.com/tarantool/tarantool/tree/HustonMmmavr/gh-4660-fix-fio-mktree-race
issue: https://github.com/tarantool/tarantool/issues/4660

 src/lua/fio.lua | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/lua/fio.lua b/src/lua/fio.lua
index cb224f3d0..c9aeac951 100644
--- a/src/lua/fio.lua
+++ b/src/lua/fio.lua
@@ -364,7 +364,7 @@ fio.mktree = function(path, mode)
         local stat = fio.stat(current_dir)
         if stat == nil then
             local st, err = fio.mkdir(current_dir, mode)
-            if err ~= nil  then
+            if err ~= nil and not fio.path.is_dir(current_dir) then
                 return false, string.format("Error creating directory %s: %s",
                     current_dir, tostring(err))
             end
-- 
2.23.0

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [Tarantool-patches] [PATCH] fio: fix race condition in mktree
  2019-12-17 15:52 [Tarantool-patches] [PATCH] fio: fix race condition in mktree HustonMmmavr
@ 2020-01-10 11:33 ` Igor Munkin
  0 siblings, 0 replies; 2+ messages in thread
From: Igor Munkin @ 2020-01-10 11:33 UTC (permalink / raw)
  To: HustonMmmavr; +Cc: tarantool-patches

Hello,

Thanks for the patch. Here are some general comments for the patch:
* Why this patch is necessary? I see no words in documentation about the
  concurrent path creation and corresponding errors. Therefore
  considering our docs everything works as described. If you see this as
  a gray zone, please file a ticket to tarantool/doc[1] repo for
  adjusting the fio docs regarding the changes you provide.
* If you are affected with such behaviour please describe the problem
  you faced with a bit more words in the commit message (since the word
  "parallel" is ambigous for this case) and provide a test case
  regarding your fix if possible.

There are also several comments below, please also consider them.

On 17.12.19, HustonMmmavr wrote:
> This patch fixes a bug that occours when two instances calls

Typo: s/occours/occurs/.
Typo: s/calls/call/.

> fio.mktree for the same directory in parallel.
> 
> Closes #4660
> ---
> branch: https://github.com/tarantool/tarantool/tree/HustonMmmavr/gh-4660-fix-fio-mktree-race
> issue: https://github.com/tarantool/tarantool/issues/4660
> 
>  src/lua/fio.lua | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/src/lua/fio.lua b/src/lua/fio.lua
> index cb224f3d0..c9aeac951 100644
> --- a/src/lua/fio.lua
> +++ b/src/lua/fio.lua
> @@ -364,7 +364,7 @@ fio.mktree = function(path, mode)
>          local stat = fio.stat(current_dir)
>          if stat == nil then
>              local st, err = fio.mkdir(current_dir, mode)
> -            if err ~= nil  then
> +            if err ~= nil and not fio.path.is_dir(current_dir) then

Side note: it's weird to check the second optional result value to
detect the failure instead of the first one. Feel free to ignore.

>                  return false, string.format("Error creating directory %s: %s",
>                      current_dir, tostring(err))
>              end
> -- 
> 2.23.0
> 

[1]: https://github.com/tarantool/doc

-- 
Best regards,
IM

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2020-01-10 11:35 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-12-17 15:52 [Tarantool-patches] [PATCH] fio: fix race condition in mktree HustonMmmavr
2020-01-10 11:33 ` Igor Munkin

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox