> That being said, and correct me if I'm wrong, but the only reason that I
can think of that someone would turn the mutex on is if they are doing
something that isn't thread-safe in one of their controllers/models.

mutex is on by default. DM also made some recent progress to become async, I
heard of stories/rumors about issues with the mutex being off and DM doing
some weird stuff, however it was a long time ago and it was just rumors.

It's probably safer to leave the mutex on if you don't need the speed.

-Matt


On Tue, Sep 16, 2008 at 11:55 PM, Kurt Schrader <[EMAIL PROTECTED]> wrote:
> I see.  I didn't get that the point that you were originally trying to
make is that the mutex is being wrapped around an entire request.
>
> That being said, and correct me if I'm wrong, but the only reason that I
can think of that someone would turn the mutex on is if they are doing
something that isn't thread-safe in one of their controllers/models.
>
> Presumably DM has its own conflict resolution mechanisms at the database
level.
>
> -----Original Message-----
> From: "Matt Aimonetti" <[EMAIL PROTECTED]>
>
> Date: Tue, 16 Sep 2008 23:44:04
> To: <[email protected]>
> Subject: [merb] Re: Merb mutex lock
>
>
>
> Kurt, this is exactly my point. I just wanted to show that removing
> the mutex on slow actions would obviously make the application way
> faster. The benchmark numbers aren't important.
>
> What's important to see is that when the mutex is on it's applied to
> your action and not only the DB transactions, here is the action I
> tested:
>
>  def index
>    sleep(0.1)
>    Test.first.text
>  end
>
> (Test beind a DM model with a text attribute)
>
> I hope my intentions are clearer now.
>
> -Matt
>
>
>
>
> On Tue, Sep 16, 2008 at 11:37 PM, Kurt Schrader <[EMAIL PROTECTED]>
wrote:
>> If you add in 0.1 seconds of sleep time inside a mutex then you've
rate-limited yourself to a maximum of 10 requests a second, as each request
can't process until the previous one is finished. (1/.1 = 10)
>>
>> This is not a good benchmark.
>>
>> -Kurt
>>
>> -----Original Message-----
>> From: "Matt Aimonetti" <[EMAIL PROTECTED]>
>>
>> Date: Tue, 16 Sep 2008 22:57:47
>> To: <[email protected]>
>> Cc: <[EMAIL PROTECTED]>
>> Subject: [merb] Merb mutex lock
>>
>>
>>
>> After John Bressnik from SDRuby asked questions about Merb mutex, I
>> run some benchmarks.
>>
>> I was hitting a controller making 1 simple DB request and I couldn't
>> see a difference between the mutex lock being on or off.
>>
>> I then added a sleep of 0.1 in the controller are here are the dramatic
results:
>>
>> http://gist.github.com/11201
>>
>> Requests per second:    9.81 [#/sec] (mean) with mutex on VS Requests
>> per second:    83.26 [#/sec] (mean) mutex off
>>
>> The benchmark app was using DM.
>>
>> -Matt
>>
>>
>>
>> >
>>
>
>
>
> >
>

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"merb" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at http://groups.google.com/group/merb?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to