From: "Sergey Petrenko" <sergepetrenko@tarantool.org>
To: "Vladislav Shpilevoy" <v.shpilevoy@tarantool.org>
Cc: tarantool-patches@dev.tarantool.org
Subject: Re: [Tarantool-patches] [PATCH 2/2] app/fiber: wait till a full event loop iteration ends.
Date: Thu, 14 Nov 2019 17:21:55 +0300 [thread overview]
Message-ID: <1573741315.750102299@f189.i.mail.ru> (raw)
In-Reply-To: <649a8cbc-6861-1a9f-7463-457c043a6484@tarantool.org>
Hi! Thank you for review!
>Четверг, 14 ноября 2019, 1:45 +03:00 от Vladislav Shpilevoy <v.shpilevoy@tarantool.org>:
>
>Hi! Thanks for the path!
>
>We don't use dots in commit title. I guess this is a typo.
Fixed.
>
>
>On 13/11/2019 19:04, Serge Petrenko wrote:
>> fiber.top() fills in statistics every event loop iteration,
>> so if it was just enabled, fiber.top() may contain 'inf's and
>> 'nan's in fiber cpu usage averages because total time consumed by
>> the main thread was not yet accounted for.
>> Same stands for viewing top() results for a freshly created fiber:
>> its metrics will be zero since it hasn't lived a full ev loop iteration
>> yet.
>> Fix this by delaying the test till top() results are meaningful and add
>> minor refactoring.
>>
>> Follow-up #2694
>> ---
>> test/app/fiber.result | 35 ++++++++++++++++++++++++++---------
>> test/app/fiber.test.lua | 31 +++++++++++++++++++++++--------
>> 2 files changed, 49 insertions(+), 17 deletions(-)
>>
>> diff --git a/test/app/fiber.result b/test/app/fiber.result
>> index 4a094939f..d447a36fc 100644
>> --- a/test/app/fiber.result
>> +++ b/test/app/fiber.result
>> @@ -1469,6 +1469,19 @@ sum = 0
>> fiber.top_enable()
>> ---
>> ...
>> +-- Check that a number is finite.
>> +-- Lua doesn't have this, sorry.
>> +function finite(num)\
>> + return type(num) == 'number' and num < math.huge and\
>> + num > -math.huge and tostring(num) ~= 'nan'\
>
>1. A canonical way to check for NaN is compare a value with
>itself. If they are not equal, then it is NaN.
Fixed, thanks!
>
>
>But more important questions are:
>- How can a number from top() have a not 'number' type?
It can't. I just wanted to implement a caconical is_finite check.
I can remove it, if you want me to.
>
>- How can top() contain a NaN, and an infinite value?
NaN: you issue fiber.top() on the same iteration you called
fiber.top_enable(). cord()->clock_delta_last and fiber()->clock_delta_last
both are 0, because clock_delta_last contains data from a previous ev loop
iteration. Division gives you NaN.
>
>
>> +end
>> +---
>> +...
>> +-- Wait till a full event loop iteration passes, so that
>> +-- top() contains meaningful results.
>> +while not finite(fiber.top().cpu["1/sched"].instant) do fiber.yield() end
>
>2. Double whitespace after 'do'.
Fixed, thanks.
Incremental diff follows.
diff --git a/test/app/fiber.result b/test/app/fiber.result
index d447a36fc..8eb506bd5 100644
--- a/test/app/fiber.result
+++ b/test/app/fiber.result
@@ -1469,17 +1469,14 @@ sum = 0
fiber.top_enable()
---
...
--- Check that a number is finite.
--- Lua doesn't have this, sorry.
-function finite(num)\
- return type(num) == 'number' and num < math.huge and\
- num > -math.huge and tostring(num) ~= 'nan'\
+function isnan(num)\
+ return num ~= num\
end
---
...
-- Wait till a full event loop iteration passes, so that
-- top() contains meaningful results.
-while not finite(fiber.top().cpu["1/sched"].instant) do fiber.yield() end
+while isnan(fiber.top().cpu["1/sched"].instant) do fiber.yield() end
---
...
a = fiber.top()
diff --git a/test/app/fiber.test.lua b/test/app/fiber.test.lua
index 33ebe063f..24cb8b492 100644
--- a/test/app/fiber.test.lua
+++ b/test/app/fiber.test.lua
@@ -634,16 +634,13 @@ sum = 0
-- gh-2694 fiber.top()
fiber.top_enable()
--- Check that a number is finite.
--- Lua doesn't have this, sorry.
-function finite(num)\
- return type(num) == 'number' and num < math.huge and\
- num > -math.huge and tostring(num) ~= 'nan'\
+function isnan(num)\
+ return num ~= num\
end
-- Wait till a full event loop iteration passes, so that
-- top() contains meaningful results.
-while not finite(fiber.top().cpu["1/sched"].instant) do fiber.yield() end
+while isnan(fiber.top().cpu["1/sched"].instant) do fiber.yield() end
a = fiber.top()
type(a)
Regards,
Sergey Petrenko
next prev parent reply other threads:[~2019-11-14 14:21 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-11-13 18:03 [Tarantool-patches] [PATCH 0/2] fiber.top(): minor fixup Serge Petrenko
2019-11-13 18:04 ` [Tarantool-patches] [PATCH 1/2] fiber: reset clock stats on fiber.top_enable() Serge Petrenko
2019-11-13 18:04 ` [Tarantool-patches] [PATCH 2/2] app/fiber: wait till a full event loop iteration ends Serge Petrenko
2019-11-13 22:52 ` Vladislav Shpilevoy
2019-11-14 14:21 ` Sergey Petrenko [this message]
2019-11-14 21:44 ` Vladislav Shpilevoy
2019-11-15 14:59 ` Serge Petrenko
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1573741315.750102299@f189.i.mail.ru \
--to=sergepetrenko@tarantool.org \
--cc=tarantool-patches@dev.tarantool.org \
--cc=v.shpilevoy@tarantool.org \
--subject='Re: [Tarantool-patches] [PATCH 2/2] app/fiber: wait till a full event loop iteration ends.' \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox