You are viewing a plain text version of this content. The canonical link for it is here.
Posted to olio-user@incubator.apache.org by Harold Lim <ro...@yahoo.com> on 2009/09/03 14:17:58 UTC
Question about Olio/Faban
Hi All,
Is it possible in Faban to increase the number of concurrent users in the middle of the run? If not, is it possible to run 2 Faban instance (maybe on different machines) and run the other benchmark run after the first one is running a while? Will it cause any problems on the faban agents side if 2 instances are talking to it?
I'm interested to see how the response time behaves/pattern when the number of concurrent users change.
Thanks,
Harold
Re: Question about Olio/Faban
Posted by Akara Sucharitakul <Ak...@Sun.COM>.
Again, this is the future. You can certainly parse entries out of the
log.xml in your run directory at this point.
As for the load variation, I'll see what I need to do. I cannot do a big
surgery for Faban 1.0 anymore but if it is low risk, I'll fix it.
Otherwise we'll need to do some post 1.0 build and give that to you.
-Akara
Harold Lim wrote:
> Hi Akara,
>
> Is there a ways to dump the run time report to a file? I think when I tried it, it just display it on the screen.
>
> Also, related to the previous problem, since ramp-up starts after the variable-loads, I don't see any run-time stats being reported during variable load.
>
>
> In the "ideal" case, what should the behavior be?
>
> Should it be something like this:
> ramp up start
> ramp up complete
> variable start
> variable complete
> ramp down start
> ramp down complete?
>
> Currently, what I am seeing is:
> variable start
> variable complete
> ramp up start
> ramp up complete
> steady state start
> steady state complete
> ramp down start
> ramp down complete.
>
> Since the total time for my variable load is = time for steady state.
> It takes twice as long to finish my benchmark runs.
>
>
>
>
> -Harold
>
> --- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM> wrote:
>
>> From: Akara Sucharitakul <Ak...@Sun.COM>
>> Subject: Re: Question about Olio/Faban
>> To: olio-user@incubator.apache.org
>> Date: Friday, September 18, 2009, 5:44 PM
>> BTW, it does not yet spit out a
>> graph. That's a future plan. But you can
>> easily use the data to plot a graph.
>>
>> -Akara
>>
>> Akara Sucharitakul wrote:
>>> Please see this URL for the stats it spits out:
>>> http://faban.sunsource.net/1.0/docs/howdoi/runtimereport.html
>>>
>>> Thanks,
>>> -Akara
>>>
>>> Harold Lim wrote:
>>>> I'm currently still using Faban 1.0(ea) build
>> 070209.
>>>> I'll try it out with RC1.
>>>>
>>>> What kind of stats do the runtime enabled dump
>> out?
>>>> I think I'm just interested in the response
>> time... Like for example I
>>>> have a variable load of 300 -> 400 -> 500. I
>> want to see a response
>>>> time graph, where the response time increases as I
>> increase the load.
>>>> Thanks,
>>>> -Harold
>>>>
>>>> --- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM>
>>>> wrote:
>>>>
>>>>> From: Akara Sucharitakul <Ak...@Sun.COM>
>>>>> Subject: Re: Question about Olio/Faban
>>>>> To: olio-user@incubator.apache.org
>>>>> Date: Friday, September 18, 2009, 3:44 PM
>>>>> I checked real quick for the behavior
>>>>> of the variable load and with the current
>> source, I don't think this
>>>>> behavior would be there.
>>>>> The threads get started and notifies the
>> master before it goes into the
>>>>> load variation so you should not see the log
>> sequence as you
>>>>> send me before. Can you check whether you have
>> an older Faban build?
>>>>> If so,
>>>>> can you please upgrade to 1.0RC1 (on http://faban.sunsource.net/).
>>>>>
>>>>> As for results, the final results will not
>> tell you much.
>>>>> You should rather enable the runtime stats in
>> your run confirguration
>>>>> file. Search for <runtimeStats
>> enabled="true|false">. It will
>>>>> periodically dump out the current stats into
>> the logs (I know this
>>>>> also needs
>>>>> improvement). Thanks.
>>>>>
>>>>> -Akara
>>>>>
>>>>> Harold Lim wrote:
>>>>>> I'm curious as to whether the final
>> figure/data will
>>>>> capture the variable workload or just ramp up
>> -> steady
>>>>> state -> ramp down data.
>>>>>> I'm interested in the response time graph
>> of the
>>>>> variable load.
>>>>>>
>>>>>> -Harold
>>>>>>
>>>>>>
>>>>>> --- On Fri, 9/18/09, Akara Sucharitakul
>> <Ak...@Sun.COM>
>>>>> wrote:
>>>>>>> From: Akara Sucharitakul <Ak...@Sun.COM>
>>>>>>> Subject: Re: Question about
>> Olio/Faban
>>>>>>> To: olio-user@incubator.apache.org
>>>>>>> Date: Friday, September 18, 2009, 3:16
>> PM
>>>>>>> Let me take a look at it. The feature
>>>>>>> needs a lot of work anyway. Will take
>> some time, though.
>>>>>>> -Akara
>>>>>>>
>>>>>>> Harold Lim wrote:
>>>>>>>> Hi Akara,
>>>>>>>>
>>>>>>>> I did what the doc said but I'm
>> getting a
>>>>> weird
>>>>>>> behavior.
>>>>>>>> My run.xml looks something like
>> this:
>>>>>>>> <fa:runControl>
>>>>>>>>
>> <fa:rampUp>60</fa:rampUp>
>> <fa:variableLoad>true</fa:variableLoad>
>> <fa:variableLoadFile>/root/faban/config/profiles/test/variableFile</fa:variableLoadFile>
>>
>>>>>
>> <fa:steadyState>3600</fa:steadyState>
>> <fa:rampDown>60</fa:rampDown>
>>>>>>>> </fa:runControl>
>>>>>>>>
>>>>>>>> My variableFile: 1200, 300
>>>>>>>> 1200, 400
>>>>>>>> 1200, 500
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> The behavior I'm getting is Faban
>> runs the
>>>>> variable
>>>>>>> load first, then after running the
>> variable load,
>>>>> it then
>>>>>>> starts ramping up->steady state
>> -> ramp
>>>>> down.
>>>>>>>> 10:28:25
>> shirako067
>>>>>>> INFO
>>>>> Variable load
>>>>>>> controller thread started!
>>>>>>>> 10:28:25
>> shirako067
>>>>>>> INFO
>>>>> Variable load
>>>>>>> controller thread advanced!
>>>>>>>> 10:28:25
>> shirako067
>>>>>>> INFO
>>>>> Variable load
>>>>>>> controller thread sleeping for 1200
>> seconds.
>>>>> Setting for 300
>>>>>>> active threads.
>>>>>>>> 10:48:25
>> shirako067
>>>>>>> INFO
>>>>> Variable load
>>>>>>> controller thread advanced!
>>>>>>>> 10:48:25
>> shirako067
>>>>>>> INFO
>>>>> Variable load
>>>>>>> controller thread sleeping for 1200
>> seconds.
>>>>> Setting for 400
>>>>>>> active threads
>>>>>>>> 11:08:25
>> shirako067
>>>>>>> INFO
>>>>> Variable load
>>>>>>> controller thread advanced!
>>>>>>>> 11:08:25
>> shirako067
>>>>>>> INFO
>>>>> Variable load
>>>>>>> controller thread sleeping for 1200
>> seconds.
>>>>> Setting for 500
>>>>>>> active threads.
>>>>>>>> 11:28:25
>> shirako067
>>>>>>> INFO
>>>>> Started all
>>>>>>> threads; run commences in 2964 ms
>>>>>>>> 11:28:29
>>
>>>>>>> INFO Ramp up
>> started
>>>>>>>> 11:29:28
>> shirako067
>>>>>>> INFO
>>>>> Ramp up
>>>>>>> completed
>>>>>>>> ...
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> -Harold
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> --- On Thu, 9/3/09, Akara
>> Sucharitakul <Ak...@Sun.COM>
>>>>>>> wrote:
>>>>>>>>> From: Akara Sucharitakul
>> <Ak...@Sun.COM>
>>>>>>>>> Subject: Re: Question about
>> Olio/Faban
>>>>>>>>> To: olio-user@incubator.apache.org
>>>>>>>>> Date: Thursday, September 3,
>> 2009, 12:54
>>>>> PM
>>>>>>>>> Yes, but the feature is still
>> quite
>>>>>>>>> restrictive. We're building it
>> out better
>>>>> in the
>>>>>>> future.
>>>>>>>>> Please see how to do it here.
>> Let us know
>>>>> if you
>>>>>>> run into
>>>>>>>>> problems.
>>>>>>>>>
>>>>>>>>> http://faban.sunsource.net/1.0/docs/howdoi/loadvariation.html
>>>>>>>>>
>>>>>>>>> Adding yet another Faban
>> instance
>>>>> altogether will
>>>>>>> cause
>>>>>>>>> conflicts.
>>>>>>>>>
>>>>>>>>> -Akara
>>>>>>>>>
>>>>>>>>> Harold Lim wrote:
>>>>>>>>>> Hi All,
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Is it possible in Faban to
>> increase
>>>>> the number
>>>>>>> of
>>>>>>>>> concurrent users in the middle
>> of the run?
>>>>> If not,
>>>>>>> is it
>>>>>>>>> possible to run 2 Faban
>> instance (maybe
>>>>> on
>>>>>>> different
>>>>>>>>> machines) and run the other
>> benchmark run
>>>>> after
>>>>>>> the first
>>>>>>>>> one is running a while? Will
>> it cause any
>>>>> problems
>>>>>>> on
>>>>>>>>> the faban agents side if
>> 2 instances
>>>>> are
>>>>>>> talking to
>>>>>>>>> it?
>>>>>>>>>> I'm interested to see how
>> the response
>>>>> time
>>>>>>>>> behaves/pattern when the
>> number of
>>>>> concurrent
>>>>>>> users change.
>>>>>>>>>> Thanks,
>>>>>>>>>> Harold
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>
>>>>>>
>>>>
>>>>
>>
>
>
>
Re: Question about Olio/Faban
Posted by Harold Lim <ro...@yahoo.com>.
Hi Akara,
Is there a ways to dump the run time report to a file? I think when I tried it, it just display it on the screen.
Also, related to the previous problem, since ramp-up starts after the variable-loads, I don't see any run-time stats being reported during variable load.
In the "ideal" case, what should the behavior be?
Should it be something like this:
ramp up start
ramp up complete
variable start
variable complete
ramp down start
ramp down complete?
Currently, what I am seeing is:
variable start
variable complete
ramp up start
ramp up complete
steady state start
steady state complete
ramp down start
ramp down complete.
Since the total time for my variable load is = time for steady state.
It takes twice as long to finish my benchmark runs.
-Harold
--- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM> wrote:
> From: Akara Sucharitakul <Ak...@Sun.COM>
> Subject: Re: Question about Olio/Faban
> To: olio-user@incubator.apache.org
> Date: Friday, September 18, 2009, 5:44 PM
> BTW, it does not yet spit out a
> graph. That's a future plan. But you can
> easily use the data to plot a graph.
>
> -Akara
>
> Akara Sucharitakul wrote:
> > Please see this URL for the stats it spits out:
> > http://faban.sunsource.net/1.0/docs/howdoi/runtimereport.html
> >
> > Thanks,
> > -Akara
> >
> > Harold Lim wrote:
> >> I'm currently still using Faban 1.0(ea) build
> 070209.
> >> I'll try it out with RC1.
> >>
> >> What kind of stats do the runtime enabled dump
> out?
> >>
> >> I think I'm just interested in the response
> time... Like for example I
> >> have a variable load of 300 -> 400 -> 500. I
> want to see a response
> >> time graph, where the response time increases as I
> increase the load.
> >>
> >> Thanks,
> >> -Harold
> >>
> >> --- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM>
>
> >> wrote:
> >>
> >>> From: Akara Sucharitakul <Ak...@Sun.COM>
> >>> Subject: Re: Question about Olio/Faban
> >>> To: olio-user@incubator.apache.org
> >>> Date: Friday, September 18, 2009, 3:44 PM
> >>> I checked real quick for the behavior
> >>> of the variable load and with the current
> source, I don't think this
> >>> behavior would be there.
> >>> The threads get started and notifies the
> master before it goes into the
> >>> load variation so you should not see the log
> sequence as you
> >>> send me before. Can you check whether you have
> an older Faban build?
> >>> If so,
> >>> can you please upgrade to 1.0RC1 (on http://faban.sunsource.net/).
> >>>
> >>> As for results, the final results will not
> tell you much.
> >>> You should rather enable the runtime stats in
> your run confirguration
> >>> file. Search for <runtimeStats
> enabled="true|false">. It will
> >>> periodically dump out the current stats into
> the logs (I know this
> >>> also needs
> >>> improvement). Thanks.
> >>>
> >>> -Akara
> >>>
> >>> Harold Lim wrote:
> >>>> I'm curious as to whether the final
> figure/data will
> >>> capture the variable workload or just ramp up
> -> steady
> >>> state -> ramp down data.
> >>>> I'm interested in the response time graph
> of the
> >>> variable load.
> >>>>
> >>>>
> >>>> -Harold
> >>>>
> >>>>
> >>>> --- On Fri, 9/18/09, Akara Sucharitakul
> <Ak...@Sun.COM>
> >>> wrote:
> >>>>> From: Akara Sucharitakul <Ak...@Sun.COM>
> >>>>> Subject: Re: Question about
> Olio/Faban
> >>>>> To: olio-user@incubator.apache.org
> >>>>> Date: Friday, September 18, 2009, 3:16
> PM
> >>>>> Let me take a look at it. The feature
> >>>>> needs a lot of work anyway. Will take
> some time, though.
> >>>>>
> >>>>> -Akara
> >>>>>
> >>>>> Harold Lim wrote:
> >>>>>> Hi Akara,
> >>>>>>
> >>>>>> I did what the doc said but I'm
> getting a
> >>> weird
> >>>>> behavior.
> >>>>>> My run.xml looks something like
> this:
> >>>>>>
> >>>>>> <fa:runControl>
> >>>>>>
> <fa:rampUp>60</fa:rampUp>
> >>>>>>
> >>>
> <fa:variableLoad>true</fa:variableLoad>
> >>>
> <fa:variableLoadFile>/root/faban/config/profiles/test/variableFile</fa:variableLoadFile>
>
> >>>
> >>>
> <fa:steadyState>3600</fa:steadyState>
> >>>>>>
> <fa:rampDown>60</fa:rampDown>
> >>>>>> </fa:runControl>
> >>>>>>
> >>>>>> My variableFile: 1200, 300
> >>>>>> 1200, 400
> >>>>>> 1200, 500
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> The behavior I'm getting is Faban
> runs the
> >>> variable
> >>>>> load first, then after running the
> variable load,
> >>> it then
> >>>>> starts ramping up->steady state
> -> ramp
> >>> down.
> >>>>>> 10:28:25
> shirako067
> >>>>> INFO
> >>> Variable load
> >>>>> controller thread started!
> >>>>>> 10:28:25
> shirako067
> >>>>> INFO
> >>> Variable load
> >>>>> controller thread advanced!
> >>>>>> 10:28:25
> shirako067
> >>>>> INFO
> >>> Variable load
> >>>>> controller thread sleeping for 1200
> seconds.
> >>> Setting for 300
> >>>>> active threads.
> >>>>>> 10:48:25
> shirako067
> >>>>> INFO
> >>> Variable load
> >>>>> controller thread advanced!
> >>>>>> 10:48:25
> shirako067
> >>>>> INFO
> >>> Variable load
> >>>>> controller thread sleeping for 1200
> seconds.
> >>> Setting for 400
> >>>>> active threads
> >>>>>> 11:08:25
> shirako067
> >>>>> INFO
> >>> Variable load
> >>>>> controller thread advanced!
> >>>>>> 11:08:25
> shirako067
> >>>>> INFO
> >>> Variable load
> >>>>> controller thread sleeping for 1200
> seconds.
> >>> Setting for 500
> >>>>> active threads.
> >>>>>> 11:28:25
> shirako067
> >>>>> INFO
> >>> Started all
> >>>>> threads; run commences in 2964 ms
> >>>>>> 11:28:29
>
> >>>>> INFO Ramp up
> started
> >>>>>> 11:29:28
> shirako067
> >>>>> INFO
> >>> Ramp up
> >>>>> completed
> >>>>>> ...
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> -Harold
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> --- On Thu, 9/3/09, Akara
> Sucharitakul <Ak...@Sun.COM>
> >>>>> wrote:
> >>>>>>> From: Akara Sucharitakul
> <Ak...@Sun.COM>
> >>>>>>> Subject: Re: Question about
> Olio/Faban
> >>>>>>> To: olio-user@incubator.apache.org
> >>>>>>> Date: Thursday, September 3,
> 2009, 12:54
> >>> PM
> >>>>>>> Yes, but the feature is still
> quite
> >>>>>>> restrictive. We're building it
> out better
> >>> in the
> >>>>> future.
> >>>>>>> Please see how to do it here.
> Let us know
> >>> if you
> >>>>> run into
> >>>>>>> problems.
> >>>>>>>
> >>>>>>> http://faban.sunsource.net/1.0/docs/howdoi/loadvariation.html
> >>>>>>>
> >>>>>>> Adding yet another Faban
> instance
> >>> altogether will
> >>>>> cause
> >>>>>>> conflicts.
> >>>>>>>
> >>>>>>> -Akara
> >>>>>>>
> >>>>>>> Harold Lim wrote:
> >>>>>>>> Hi All,
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> Is it possible in Faban to
> increase
> >>> the number
> >>>>> of
> >>>>>>> concurrent users in the middle
> of the run?
> >>> If not,
> >>>>> is it
> >>>>>>> possible to run 2 Faban
> instance (maybe
> >>> on
> >>>>> different
> >>>>>>> machines) and run the other
> benchmark run
> >>> after
> >>>>> the first
> >>>>>>> one is running a while? Will
> it cause any
> >>> problems
> >>>>> on
> >>>>>>> the faban agents side if
> 2 instances
> >>> are
> >>>>> talking to
> >>>>>>> it?
> >>>>>>>> I'm interested to see how
> the response
> >>> time
> >>>>>>> behaves/pattern when the
> number of
> >>> concurrent
> >>>>> users change.
> >>>>>>>> Thanks,
> >>>>>>>> Harold
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>>
> >>>>>>
> >>>>
> >>>>
> >>>
> >>
> >>
> >>
> >
>
>
Re: Question about Olio/Faban
Posted by Akara Sucharitakul <Ak...@Sun.COM>.
BTW, it does not yet spit out a graph. That's a future plan. But you can
easily use the data to plot a graph.
-Akara
Akara Sucharitakul wrote:
> Please see this URL for the stats it spits out:
> http://faban.sunsource.net/1.0/docs/howdoi/runtimereport.html
>
> Thanks,
> -Akara
>
> Harold Lim wrote:
>> I'm currently still using Faban 1.0(ea) build 070209.
>> I'll try it out with RC1.
>>
>> What kind of stats do the runtime enabled dump out?
>>
>> I think I'm just interested in the response time... Like for example I
>> have a variable load of 300 -> 400 -> 500. I want to see a response
>> time graph, where the response time increases as I increase the load.
>>
>> Thanks,
>> -Harold
>>
>> --- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM>
>> wrote:
>>
>>> From: Akara Sucharitakul <Ak...@Sun.COM>
>>> Subject: Re: Question about Olio/Faban
>>> To: olio-user@incubator.apache.org
>>> Date: Friday, September 18, 2009, 3:44 PM
>>> I checked real quick for the behavior
>>> of the variable load and with the current source, I don't think this
>>> behavior would be there.
>>> The threads get started and notifies the master before it goes into the
>>> load variation so you should not see the log sequence as you
>>> send me before. Can you check whether you have an older Faban build?
>>> If so,
>>> can you please upgrade to 1.0RC1 (on http://faban.sunsource.net/).
>>>
>>> As for results, the final results will not tell you much.
>>> You should rather enable the runtime stats in your run confirguration
>>> file. Search for <runtimeStats enabled="true|false">. It will
>>> periodically dump out the current stats into the logs (I know this
>>> also needs
>>> improvement). Thanks.
>>>
>>> -Akara
>>>
>>> Harold Lim wrote:
>>>> I'm curious as to whether the final figure/data will
>>> capture the variable workload or just ramp up -> steady
>>> state -> ramp down data.
>>>> I'm interested in the response time graph of the
>>> variable load.
>>>>
>>>>
>>>> -Harold
>>>>
>>>>
>>>> --- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM>
>>> wrote:
>>>>> From: Akara Sucharitakul <Ak...@Sun.COM>
>>>>> Subject: Re: Question about Olio/Faban
>>>>> To: olio-user@incubator.apache.org
>>>>> Date: Friday, September 18, 2009, 3:16 PM
>>>>> Let me take a look at it. The feature
>>>>> needs a lot of work anyway. Will take some time, though.
>>>>>
>>>>> -Akara
>>>>>
>>>>> Harold Lim wrote:
>>>>>> Hi Akara,
>>>>>>
>>>>>> I did what the doc said but I'm getting a
>>> weird
>>>>> behavior.
>>>>>> My run.xml looks something like this:
>>>>>>
>>>>>> <fa:runControl>
>>>>>> <fa:rampUp>60</fa:rampUp>
>>>>>>
>>> <fa:variableLoad>true</fa:variableLoad>
>>> <fa:variableLoadFile>/root/faban/config/profiles/test/variableFile</fa:variableLoadFile>
>>>
>>> <fa:steadyState>3600</fa:steadyState>
>>>>>> <fa:rampDown>60</fa:rampDown>
>>>>>> </fa:runControl>
>>>>>>
>>>>>> My variableFile: 1200, 300
>>>>>> 1200, 400
>>>>>> 1200, 500
>>>>>>
>>>>>>
>>>>>>
>>>>>> The behavior I'm getting is Faban runs the
>>> variable
>>>>> load first, then after running the variable load,
>>> it then
>>>>> starts ramping up->steady state -> ramp
>>> down.
>>>>>> 10:28:25 shirako067
>>>>> INFO
>>> Variable load
>>>>> controller thread started!
>>>>>> 10:28:25 shirako067
>>>>> INFO
>>> Variable load
>>>>> controller thread advanced!
>>>>>> 10:28:25 shirako067
>>>>> INFO
>>> Variable load
>>>>> controller thread sleeping for 1200 seconds.
>>> Setting for 300
>>>>> active threads.
>>>>>> 10:48:25 shirako067
>>>>> INFO
>>> Variable load
>>>>> controller thread advanced!
>>>>>> 10:48:25 shirako067
>>>>> INFO
>>> Variable load
>>>>> controller thread sleeping for 1200 seconds.
>>> Setting for 400
>>>>> active threads
>>>>>> 11:08:25 shirako067
>>>>> INFO
>>> Variable load
>>>>> controller thread advanced!
>>>>>> 11:08:25 shirako067
>>>>> INFO
>>> Variable load
>>>>> controller thread sleeping for 1200 seconds.
>>> Setting for 500
>>>>> active threads.
>>>>>> 11:28:25 shirako067
>>>>> INFO
>>> Started all
>>>>> threads; run commences in 2964 ms
>>>>>> 11:28:29
>>>>> INFO Ramp up started
>>>>>> 11:29:28 shirako067
>>>>> INFO
>>> Ramp up
>>>>> completed
>>>>>> ...
>>>>>>
>>>>>>
>>>>>>
>>>>>> -Harold
>>>>>>
>>>>>>
>>>>>>
>>>>>> --- On Thu, 9/3/09, Akara Sucharitakul <Ak...@Sun.COM>
>>>>> wrote:
>>>>>>> From: Akara Sucharitakul <Ak...@Sun.COM>
>>>>>>> Subject: Re: Question about Olio/Faban
>>>>>>> To: olio-user@incubator.apache.org
>>>>>>> Date: Thursday, September 3, 2009, 12:54
>>> PM
>>>>>>> Yes, but the feature is still quite
>>>>>>> restrictive. We're building it out better
>>> in the
>>>>> future.
>>>>>>> Please see how to do it here. Let us know
>>> if you
>>>>> run into
>>>>>>> problems.
>>>>>>>
>>>>>>> http://faban.sunsource.net/1.0/docs/howdoi/loadvariation.html
>>>>>>>
>>>>>>> Adding yet another Faban instance
>>> altogether will
>>>>> cause
>>>>>>> conflicts.
>>>>>>>
>>>>>>> -Akara
>>>>>>>
>>>>>>> Harold Lim wrote:
>>>>>>>> Hi All,
>>>>>>>>
>>>>>>>>
>>>>>>>> Is it possible in Faban to increase
>>> the number
>>>>> of
>>>>>>> concurrent users in the middle of the run?
>>> If not,
>>>>> is it
>>>>>>> possible to run 2 Faban instance (maybe
>>> on
>>>>> different
>>>>>>> machines) and run the other benchmark run
>>> after
>>>>> the first
>>>>>>> one is running a while? Will it cause any
>>> problems
>>>>> on
>>>>>>> the faban agents side if 2 instances
>>> are
>>>>> talking to
>>>>>>> it?
>>>>>>>> I'm interested to see how the response
>>> time
>>>>>>> behaves/pattern when the number of
>>> concurrent
>>>>> users change.
>>>>>>>> Thanks,
>>>>>>>> Harold
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>
>>>>>>
>>>>
>>>>
>>>
>>
>>
>>
>
Re: Question about Olio/Faban
Posted by Akara Sucharitakul <Ak...@Sun.COM>.
Please see this URL for the stats it spits out:
http://faban.sunsource.net/1.0/docs/howdoi/runtimereport.html
Thanks,
-Akara
Harold Lim wrote:
> I'm currently still using Faban 1.0(ea) build 070209.
> I'll try it out with RC1.
>
> What kind of stats do the runtime enabled dump out?
>
> I think I'm just interested in the response time... Like for example I have a variable load of 300 -> 400 -> 500. I want to see a response time graph, where the response time increases as I increase the load.
>
> Thanks,
> -Harold
>
> --- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM> wrote:
>
>> From: Akara Sucharitakul <Ak...@Sun.COM>
>> Subject: Re: Question about Olio/Faban
>> To: olio-user@incubator.apache.org
>> Date: Friday, September 18, 2009, 3:44 PM
>> I checked real quick for the behavior
>> of the variable load and with the
>> current source, I don't think this behavior would be there.
>> The threads
>> get started and notifies the master before it goes into the
>> load
>> variation so you should not see the log sequence as you
>> send me before.
>> Can you check whether you have an older Faban build? If so,
>> can you
>> please upgrade to 1.0RC1 (on http://faban.sunsource.net/).
>>
>> As for results, the final results will not tell you much.
>> You should
>> rather enable the runtime stats in your run confirguration
>> file. Search
>> for <runtimeStats enabled="true|false">. It will
>> periodically dump out
>> the current stats into the logs (I know this also needs
>> improvement).
>> Thanks.
>>
>> -Akara
>>
>> Harold Lim wrote:
>>> I'm curious as to whether the final figure/data will
>> capture the variable workload or just ramp up -> steady
>> state -> ramp down data.
>>> I'm interested in the response time graph of the
>> variable load.
>>>
>>>
>>> -Harold
>>>
>>>
>>> --- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM>
>> wrote:
>>>> From: Akara Sucharitakul <Ak...@Sun.COM>
>>>> Subject: Re: Question about Olio/Faban
>>>> To: olio-user@incubator.apache.org
>>>> Date: Friday, September 18, 2009, 3:16 PM
>>>> Let me take a look at it. The feature
>>>> needs a lot of work anyway. Will
>>>> take some time, though.
>>>>
>>>> -Akara
>>>>
>>>> Harold Lim wrote:
>>>>> Hi Akara,
>>>>>
>>>>> I did what the doc said but I'm getting a
>> weird
>>>> behavior.
>>>>> My run.xml looks something like this:
>>>>>
>>>>> <fa:runControl>
>>>>> <fa:rampUp>60</fa:rampUp>
>>>>>
>> <fa:variableLoad>true</fa:variableLoad>
>> <fa:variableLoadFile>/root/faban/config/profiles/test/variableFile</fa:variableLoadFile>
>> <fa:steadyState>3600</fa:steadyState>
>>>>> <fa:rampDown>60</fa:rampDown>
>>>>> </fa:runControl>
>>>>>
>>>>> My variableFile:
>>>>> 1200, 300
>>>>> 1200, 400
>>>>> 1200, 500
>>>>>
>>>>>
>>>>>
>>>>> The behavior I'm getting is Faban runs the
>> variable
>>>> load first, then after running the variable load,
>> it then
>>>> starts ramping up->steady state -> ramp
>> down.
>>>>> 10:28:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread started!
>>>>> 10:28:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread advanced!
>>>>> 10:28:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread sleeping for 1200 seconds.
>> Setting for 300
>>>> active threads.
>>>>> 10:48:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread advanced!
>>>>> 10:48:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread sleeping for 1200 seconds.
>> Setting for 400
>>>> active threads
>>>>> 11:08:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread advanced!
>>>>> 11:08:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread sleeping for 1200 seconds.
>> Setting for 500
>>>> active threads.
>>>>> 11:28:25 shirako067
>>>> INFO
>> Started all
>>>> threads; run commences in 2964 ms
>>>>> 11:28:29
>>>> INFO Ramp up started
>>>>> 11:29:28 shirako067
>>>> INFO
>> Ramp up
>>>> completed
>>>>> ...
>>>>>
>>>>>
>>>>>
>>>>> -Harold
>>>>>
>>>>>
>>>>>
>>>>> --- On Thu, 9/3/09, Akara Sucharitakul <Ak...@Sun.COM>
>>>> wrote:
>>>>>> From: Akara Sucharitakul <Ak...@Sun.COM>
>>>>>> Subject: Re: Question about Olio/Faban
>>>>>> To: olio-user@incubator.apache.org
>>>>>> Date: Thursday, September 3, 2009, 12:54
>> PM
>>>>>> Yes, but the feature is still quite
>>>>>> restrictive. We're building it out better
>> in the
>>>> future.
>>>>>> Please see how to do it here. Let us know
>> if you
>>>> run into
>>>>>> problems.
>>>>>>
>>>>>> http://faban.sunsource.net/1.0/docs/howdoi/loadvariation.html
>>>>>>
>>>>>> Adding yet another Faban instance
>> altogether will
>>>> cause
>>>>>> conflicts.
>>>>>>
>>>>>> -Akara
>>>>>>
>>>>>> Harold Lim wrote:
>>>>>>> Hi All,
>>>>>>>
>>>>>>>
>>>>>>> Is it possible in Faban to increase
>> the number
>>>> of
>>>>>> concurrent users in the middle of the run?
>> If not,
>>>> is it
>>>>>> possible to run 2 Faban instance (maybe
>> on
>>>> different
>>>>>> machines) and run the other benchmark run
>> after
>>>> the first
>>>>>> one is running a while? Will it cause any
>> problems
>>>> on
>>>>>> the faban agents side if 2 instances
>> are
>>>> talking to
>>>>>> it?
>>>>>>> I'm interested to see how the response
>> time
>>>>>> behaves/pattern when the number of
>> concurrent
>>>> users change.
>>>>>>> Thanks,
>>>>>>> Harold
>>>>>>>
>>>>>>>
>>>>>>>
>>
>>>>>
>>>
>>>
>>
>
>
>
Re: Question about Olio/Faban
Posted by Harold Lim <ro...@yahoo.com>.
I'm currently still using Faban 1.0(ea) build 070209.
I'll try it out with RC1.
What kind of stats do the runtime enabled dump out?
I think I'm just interested in the response time... Like for example I have a variable load of 300 -> 400 -> 500. I want to see a response time graph, where the response time increases as I increase the load.
Thanks,
-Harold
--- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM> wrote:
> From: Akara Sucharitakul <Ak...@Sun.COM>
> Subject: Re: Question about Olio/Faban
> To: olio-user@incubator.apache.org
> Date: Friday, September 18, 2009, 3:44 PM
> I checked real quick for the behavior
> of the variable load and with the
> current source, I don't think this behavior would be there.
> The threads
> get started and notifies the master before it goes into the
> load
> variation so you should not see the log sequence as you
> send me before.
> Can you check whether you have an older Faban build? If so,
> can you
> please upgrade to 1.0RC1 (on http://faban.sunsource.net/).
>
> As for results, the final results will not tell you much.
> You should
> rather enable the runtime stats in your run confirguration
> file. Search
> for <runtimeStats enabled="true|false">. It will
> periodically dump out
> the current stats into the logs (I know this also needs
> improvement).
> Thanks.
>
> -Akara
>
> Harold Lim wrote:
> > I'm curious as to whether the final figure/data will
> capture the variable workload or just ramp up -> steady
> state -> ramp down data.
> >
> > I'm interested in the response time graph of the
> variable load.
> >
> >
> >
> > -Harold
> >
> >
> > --- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM>
> wrote:
> >
> >> From: Akara Sucharitakul <Ak...@Sun.COM>
> >> Subject: Re: Question about Olio/Faban
> >> To: olio-user@incubator.apache.org
> >> Date: Friday, September 18, 2009, 3:16 PM
> >> Let me take a look at it. The feature
> >> needs a lot of work anyway. Will
> >> take some time, though.
> >>
> >> -Akara
> >>
> >> Harold Lim wrote:
> >>> Hi Akara,
> >>>
> >>> I did what the doc said but I'm getting a
> weird
> >> behavior.
> >>> My run.xml looks something like this:
> >>>
> >>> <fa:runControl>
> >>> <fa:rampUp>60</fa:rampUp>
> >>>
> <fa:variableLoad>true</fa:variableLoad>
> >>>
> >>
> <fa:variableLoadFile>/root/faban/config/profiles/test/variableFile</fa:variableLoadFile>
> >>>
> <fa:steadyState>3600</fa:steadyState>
> >>> <fa:rampDown>60</fa:rampDown>
> >>> </fa:runControl>
> >>>
> >>> My variableFile:
> >>> 1200, 300
> >>> 1200, 400
> >>> 1200, 500
> >>>
> >>>
> >>>
> >>> The behavior I'm getting is Faban runs the
> variable
> >> load first, then after running the variable load,
> it then
> >> starts ramping up->steady state -> ramp
> down.
> >>> 10:28:25 shirako067
> >> INFO
> Variable load
> >> controller thread started!
> >>> 10:28:25 shirako067
> >> INFO
> Variable load
> >> controller thread advanced!
> >>> 10:28:25 shirako067
> >> INFO
> Variable load
> >> controller thread sleeping for 1200 seconds.
> Setting for 300
> >> active threads.
> >>> 10:48:25 shirako067
> >> INFO
> Variable load
> >> controller thread advanced!
> >>> 10:48:25 shirako067
> >> INFO
> Variable load
> >> controller thread sleeping for 1200 seconds.
> Setting for 400
> >> active threads
> >>> 11:08:25 shirako067
> >> INFO
> Variable load
> >> controller thread advanced!
> >>> 11:08:25 shirako067
> >> INFO
> Variable load
> >> controller thread sleeping for 1200 seconds.
> Setting for 500
> >> active threads.
> >>> 11:28:25 shirako067
> >> INFO
> Started all
> >> threads; run commences in 2964 ms
> >>> 11:28:29
> >> INFO Ramp up started
> >>> 11:29:28 shirako067
> >> INFO
> Ramp up
> >> completed
> >>> ...
> >>>
> >>>
> >>>
> >>> -Harold
> >>>
> >>>
> >>>
> >>> --- On Thu, 9/3/09, Akara Sucharitakul <Ak...@Sun.COM>
> >> wrote:
> >>>> From: Akara Sucharitakul <Ak...@Sun.COM>
> >>>> Subject: Re: Question about Olio/Faban
> >>>> To: olio-user@incubator.apache.org
> >>>> Date: Thursday, September 3, 2009, 12:54
> PM
> >>>> Yes, but the feature is still quite
> >>>> restrictive. We're building it out better
> in the
> >> future.
> >>>> Please see how to do it here. Let us know
> if you
> >> run into
> >>>> problems.
> >>>>
> >>>> http://faban.sunsource.net/1.0/docs/howdoi/loadvariation.html
> >>>>
> >>>> Adding yet another Faban instance
> altogether will
> >> cause
> >>>> conflicts.
> >>>>
> >>>> -Akara
> >>>>
> >>>> Harold Lim wrote:
> >>>>> Hi All,
> >>>>>
> >>>>>
> >>>>> Is it possible in Faban to increase
> the number
> >> of
> >>>> concurrent users in the middle of the run?
> If not,
> >> is it
> >>>> possible to run 2 Faban instance (maybe
> on
> >> different
> >>>> machines) and run the other benchmark run
> after
> >> the first
> >>>> one is running a while? Will it cause any
> problems
> >> on
> >>>> the faban agents side if 2 instances
> are
> >> talking to
> >>>> it?
> >>>>> I'm interested to see how the response
> time
> >>>> behaves/pattern when the number of
> concurrent
> >> users change.
> >>>>> Thanks,
> >>>>> Harold
> >>>>>
> >>>>>
> >>>>>
>
> >>>
> >>>
> >>
> >
> >
> >
>
>
Re: Question about Olio/Faban
Posted by Akara Sucharitakul <Ak...@Sun.COM>.
This is odd. Certainly does not coincide with the code as I interpret
it. Something else must be holding the ramp-up back. I'll need to check.
Thanks.
-Akara
Harold Lim wrote:
> Hi Akara,
>
> I downloaded the latest Faban and I think I'm still getting that behavior.
>
> 12:51:08 shirako067 INFO Configuring 1 JavaUIDriverAgents...
> 12:59:28 shirako067 INFO JavaUIDriverAgent[0]: Successfully started 500 driver threads.
> 12:59:29 shirako067 INFO Variable load controller thread started!
> 12:59:29 shirako067 INFO Variable load controller thread advanced!
> 12:59:29 shirako067 INFO Variable load controller thread sleeping for 1200 seconds. Setting for 300 active threads.
>
> It goes to the variable load controller before the ramp-up.
>
>
> -Harold
>
>
> --- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM> wrote:
>
>> From: Akara Sucharitakul <Ak...@Sun.COM>
>> Subject: Re: Question about Olio/Faban
>> To: olio-user@incubator.apache.org
>> Date: Friday, September 18, 2009, 3:44 PM
>> I checked real quick for the behavior
>> of the variable load and with the
>> current source, I don't think this behavior would be there.
>> The threads
>> get started and notifies the master before it goes into the
>> load
>> variation so you should not see the log sequence as you
>> send me before.
>> Can you check whether you have an older Faban build? If so,
>> can you
>> please upgrade to 1.0RC1 (on http://faban.sunsource.net/).
>>
>> As for results, the final results will not tell you much.
>> You should
>> rather enable the runtime stats in your run confirguration
>> file. Search
>> for <runtimeStats enabled="true|false">. It will
>> periodically dump out
>> the current stats into the logs (I know this also needs
>> improvement).
>> Thanks.
>>
>> -Akara
>>
>> Harold Lim wrote:
>>> I'm curious as to whether the final figure/data will
>> capture the variable workload or just ramp up -> steady
>> state -> ramp down data.
>>> I'm interested in the response time graph of the
>> variable load.
>>>
>>>
>>> -Harold
>>>
>>>
>>> --- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM>
>> wrote:
>>>> From: Akara Sucharitakul <Ak...@Sun.COM>
>>>> Subject: Re: Question about Olio/Faban
>>>> To: olio-user@incubator.apache.org
>>>> Date: Friday, September 18, 2009, 3:16 PM
>>>> Let me take a look at it. The feature
>>>> needs a lot of work anyway. Will
>>>> take some time, though.
>>>>
>>>> -Akara
>>>>
>>>> Harold Lim wrote:
>>>>> Hi Akara,
>>>>>
>>>>> I did what the doc said but I'm getting a
>> weird
>>>> behavior.
>>>>> My run.xml looks something like this:
>>>>>
>>>>> <fa:runControl>
>>>>> <fa:rampUp>60</fa:rampUp>
>>>>>
>> <fa:variableLoad>true</fa:variableLoad>
>> <fa:variableLoadFile>/root/faban/config/profiles/test/variableFile</fa:variableLoadFile>
>> <fa:steadyState>3600</fa:steadyState>
>>>>> <fa:rampDown>60</fa:rampDown>
>>>>> </fa:runControl>
>>>>>
>>>>> My variableFile:
>>>>> 1200, 300
>>>>> 1200, 400
>>>>> 1200, 500
>>>>>
>>>>>
>>>>>
>>>>> The behavior I'm getting is Faban runs the
>> variable
>>>> load first, then after running the variable load,
>> it then
>>>> starts ramping up->steady state -> ramp
>> down.
>>>>> 10:28:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread started!
>>>>> 10:28:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread advanced!
>>>>> 10:28:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread sleeping for 1200 seconds.
>> Setting for 300
>>>> active threads.
>>>>> 10:48:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread advanced!
>>>>> 10:48:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread sleeping for 1200 seconds.
>> Setting for 400
>>>> active threads
>>>>> 11:08:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread advanced!
>>>>> 11:08:25 shirako067
>>>> INFO
>> Variable load
>>>> controller thread sleeping for 1200 seconds.
>> Setting for 500
>>>> active threads.
>>>>> 11:28:25 shirako067
>>>> INFO
>> Started all
>>>> threads; run commences in 2964 ms
>>>>> 11:28:29
>>>> INFO Ramp up started
>>>>> 11:29:28 shirako067
>>>> INFO
>> Ramp up
>>>> completed
>>>>> ...
>>>>>
>>>>>
>>>>>
>>>>> -Harold
>>>>>
>>>>>
>>>>>
>>>>> --- On Thu, 9/3/09, Akara Sucharitakul <Ak...@Sun.COM>
>>>> wrote:
>>>>>> From: Akara Sucharitakul <Ak...@Sun.COM>
>>>>>> Subject: Re: Question about Olio/Faban
>>>>>> To: olio-user@incubator.apache.org
>>>>>> Date: Thursday, September 3, 2009, 12:54
>> PM
>>>>>> Yes, but the feature is still quite
>>>>>> restrictive. We're building it out better
>> in the
>>>> future.
>>>>>> Please see how to do it here. Let us know
>> if you
>>>> run into
>>>>>> problems.
>>>>>>
>>>>>> http://faban.sunsource.net/1.0/docs/howdoi/loadvariation.html
>>>>>>
>>>>>> Adding yet another Faban instance
>> altogether will
>>>> cause
>>>>>> conflicts.
>>>>>>
>>>>>> -Akara
>>>>>>
>>>>>> Harold Lim wrote:
>>>>>>> Hi All,
>>>>>>>
>>>>>>>
>>>>>>> Is it possible in Faban to increase
>> the number
>>>> of
>>>>>> concurrent users in the middle of the run?
>> If not,
>>>> is it
>>>>>> possible to run 2 Faban instance (maybe
>> on
>>>> different
>>>>>> machines) and run the other benchmark run
>> after
>>>> the first
>>>>>> one is running a while? Will it cause any
>> problems
>>>> on
>>>>>> the faban agents side if 2 instances
>> are
>>>> talking to
>>>>>> it?
>>>>>>> I'm interested to see how the response
>> time
>>>>>> behaves/pattern when the number of
>> concurrent
>>>> users change.
>>>>>>> Thanks,
>>>>>>> Harold
>>>>>>>
>>>>>>>
>>>>>>>
>>
>>>>>
>>>
>>>
>>
>
>
>
Re: Question about Olio/Faban
Posted by Harold Lim <ro...@yahoo.com>.
Hi Akara,
I downloaded the latest Faban and I think I'm still getting that behavior.
12:51:08 shirako067 INFO Configuring 1 JavaUIDriverAgents...
12:59:28 shirako067 INFO JavaUIDriverAgent[0]: Successfully started 500 driver threads.
12:59:29 shirako067 INFO Variable load controller thread started!
12:59:29 shirako067 INFO Variable load controller thread advanced!
12:59:29 shirako067 INFO Variable load controller thread sleeping for 1200 seconds. Setting for 300 active threads.
It goes to the variable load controller before the ramp-up.
-Harold
--- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM> wrote:
> From: Akara Sucharitakul <Ak...@Sun.COM>
> Subject: Re: Question about Olio/Faban
> To: olio-user@incubator.apache.org
> Date: Friday, September 18, 2009, 3:44 PM
> I checked real quick for the behavior
> of the variable load and with the
> current source, I don't think this behavior would be there.
> The threads
> get started and notifies the master before it goes into the
> load
> variation so you should not see the log sequence as you
> send me before.
> Can you check whether you have an older Faban build? If so,
> can you
> please upgrade to 1.0RC1 (on http://faban.sunsource.net/).
>
> As for results, the final results will not tell you much.
> You should
> rather enable the runtime stats in your run confirguration
> file. Search
> for <runtimeStats enabled="true|false">. It will
> periodically dump out
> the current stats into the logs (I know this also needs
> improvement).
> Thanks.
>
> -Akara
>
> Harold Lim wrote:
> > I'm curious as to whether the final figure/data will
> capture the variable workload or just ramp up -> steady
> state -> ramp down data.
> >
> > I'm interested in the response time graph of the
> variable load.
> >
> >
> >
> > -Harold
> >
> >
> > --- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM>
> wrote:
> >
> >> From: Akara Sucharitakul <Ak...@Sun.COM>
> >> Subject: Re: Question about Olio/Faban
> >> To: olio-user@incubator.apache.org
> >> Date: Friday, September 18, 2009, 3:16 PM
> >> Let me take a look at it. The feature
> >> needs a lot of work anyway. Will
> >> take some time, though.
> >>
> >> -Akara
> >>
> >> Harold Lim wrote:
> >>> Hi Akara,
> >>>
> >>> I did what the doc said but I'm getting a
> weird
> >> behavior.
> >>> My run.xml looks something like this:
> >>>
> >>> <fa:runControl>
> >>> <fa:rampUp>60</fa:rampUp>
> >>>
> <fa:variableLoad>true</fa:variableLoad>
> >>>
> >>
> <fa:variableLoadFile>/root/faban/config/profiles/test/variableFile</fa:variableLoadFile>
> >>>
> <fa:steadyState>3600</fa:steadyState>
> >>> <fa:rampDown>60</fa:rampDown>
> >>> </fa:runControl>
> >>>
> >>> My variableFile:
> >>> 1200, 300
> >>> 1200, 400
> >>> 1200, 500
> >>>
> >>>
> >>>
> >>> The behavior I'm getting is Faban runs the
> variable
> >> load first, then after running the variable load,
> it then
> >> starts ramping up->steady state -> ramp
> down.
> >>> 10:28:25 shirako067
> >> INFO
> Variable load
> >> controller thread started!
> >>> 10:28:25 shirako067
> >> INFO
> Variable load
> >> controller thread advanced!
> >>> 10:28:25 shirako067
> >> INFO
> Variable load
> >> controller thread sleeping for 1200 seconds.
> Setting for 300
> >> active threads.
> >>> 10:48:25 shirako067
> >> INFO
> Variable load
> >> controller thread advanced!
> >>> 10:48:25 shirako067
> >> INFO
> Variable load
> >> controller thread sleeping for 1200 seconds.
> Setting for 400
> >> active threads
> >>> 11:08:25 shirako067
> >> INFO
> Variable load
> >> controller thread advanced!
> >>> 11:08:25 shirako067
> >> INFO
> Variable load
> >> controller thread sleeping for 1200 seconds.
> Setting for 500
> >> active threads.
> >>> 11:28:25 shirako067
> >> INFO
> Started all
> >> threads; run commences in 2964 ms
> >>> 11:28:29
> >> INFO Ramp up started
> >>> 11:29:28 shirako067
> >> INFO
> Ramp up
> >> completed
> >>> ...
> >>>
> >>>
> >>>
> >>> -Harold
> >>>
> >>>
> >>>
> >>> --- On Thu, 9/3/09, Akara Sucharitakul <Ak...@Sun.COM>
> >> wrote:
> >>>> From: Akara Sucharitakul <Ak...@Sun.COM>
> >>>> Subject: Re: Question about Olio/Faban
> >>>> To: olio-user@incubator.apache.org
> >>>> Date: Thursday, September 3, 2009, 12:54
> PM
> >>>> Yes, but the feature is still quite
> >>>> restrictive. We're building it out better
> in the
> >> future.
> >>>> Please see how to do it here. Let us know
> if you
> >> run into
> >>>> problems.
> >>>>
> >>>> http://faban.sunsource.net/1.0/docs/howdoi/loadvariation.html
> >>>>
> >>>> Adding yet another Faban instance
> altogether will
> >> cause
> >>>> conflicts.
> >>>>
> >>>> -Akara
> >>>>
> >>>> Harold Lim wrote:
> >>>>> Hi All,
> >>>>>
> >>>>>
> >>>>> Is it possible in Faban to increase
> the number
> >> of
> >>>> concurrent users in the middle of the run?
> If not,
> >> is it
> >>>> possible to run 2 Faban instance (maybe
> on
> >> different
> >>>> machines) and run the other benchmark run
> after
> >> the first
> >>>> one is running a while? Will it cause any
> problems
> >> on
> >>>> the faban agents side if 2 instances
> are
> >> talking to
> >>>> it?
> >>>>> I'm interested to see how the response
> time
> >>>> behaves/pattern when the number of
> concurrent
> >> users change.
> >>>>> Thanks,
> >>>>> Harold
> >>>>>
> >>>>>
> >>>>>
>
> >>>
> >>>
> >>
> >
> >
> >
>
>
Re: Question about Olio/Faban
Posted by Akara Sucharitakul <Ak...@Sun.COM>.
I checked real quick for the behavior of the variable load and with the
current source, I don't think this behavior would be there. The threads
get started and notifies the master before it goes into the load
variation so you should not see the log sequence as you send me before.
Can you check whether you have an older Faban build? If so, can you
please upgrade to 1.0RC1 (on http://faban.sunsource.net/).
As for results, the final results will not tell you much. You should
rather enable the runtime stats in your run confirguration file. Search
for <runtimeStats enabled="true|false">. It will periodically dump out
the current stats into the logs (I know this also needs improvement).
Thanks.
-Akara
Harold Lim wrote:
> I'm curious as to whether the final figure/data will capture the variable workload or just ramp up -> steady state -> ramp down data.
>
> I'm interested in the response time graph of the variable load.
>
>
>
> -Harold
>
>
> --- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM> wrote:
>
>> From: Akara Sucharitakul <Ak...@Sun.COM>
>> Subject: Re: Question about Olio/Faban
>> To: olio-user@incubator.apache.org
>> Date: Friday, September 18, 2009, 3:16 PM
>> Let me take a look at it. The feature
>> needs a lot of work anyway. Will
>> take some time, though.
>>
>> -Akara
>>
>> Harold Lim wrote:
>>> Hi Akara,
>>>
>>> I did what the doc said but I'm getting a weird
>> behavior.
>>> My run.xml looks something like this:
>>>
>>> <fa:runControl>
>>> <fa:rampUp>60</fa:rampUp>
>>> <fa:variableLoad>true</fa:variableLoad>
>>>
>> <fa:variableLoadFile>/root/faban/config/profiles/test/variableFile</fa:variableLoadFile>
>>> <fa:steadyState>3600</fa:steadyState>
>>> <fa:rampDown>60</fa:rampDown>
>>> </fa:runControl>
>>>
>>> My variableFile:
>>> 1200, 300
>>> 1200, 400
>>> 1200, 500
>>>
>>>
>>>
>>> The behavior I'm getting is Faban runs the variable
>> load first, then after running the variable load, it then
>> starts ramping up->steady state -> ramp down.
>>> 10:28:25 shirako067
>> INFO Variable load
>> controller thread started!
>>> 10:28:25 shirako067
>> INFO Variable load
>> controller thread advanced!
>>> 10:28:25 shirako067
>> INFO Variable load
>> controller thread sleeping for 1200 seconds. Setting for 300
>> active threads.
>>> 10:48:25 shirako067
>> INFO Variable load
>> controller thread advanced!
>>> 10:48:25 shirako067
>> INFO Variable load
>> controller thread sleeping for 1200 seconds. Setting for 400
>> active threads
>>> 11:08:25 shirako067
>> INFO Variable load
>> controller thread advanced!
>>> 11:08:25 shirako067
>> INFO Variable load
>> controller thread sleeping for 1200 seconds. Setting for 500
>> active threads.
>>> 11:28:25 shirako067
>> INFO Started all
>> threads; run commences in 2964 ms
>>> 11:28:29
>> INFO Ramp up started
>>> 11:29:28 shirako067
>> INFO Ramp up
>> completed
>>> ...
>>>
>>>
>>>
>>> -Harold
>>>
>>>
>>>
>>> --- On Thu, 9/3/09, Akara Sucharitakul <Ak...@Sun.COM>
>> wrote:
>>>> From: Akara Sucharitakul <Ak...@Sun.COM>
>>>> Subject: Re: Question about Olio/Faban
>>>> To: olio-user@incubator.apache.org
>>>> Date: Thursday, September 3, 2009, 12:54 PM
>>>> Yes, but the feature is still quite
>>>> restrictive. We're building it out better in the
>> future.
>>>> Please see how to do it here. Let us know if you
>> run into
>>>> problems.
>>>>
>>>> http://faban.sunsource.net/1.0/docs/howdoi/loadvariation.html
>>>>
>>>> Adding yet another Faban instance altogether will
>> cause
>>>> conflicts.
>>>>
>>>> -Akara
>>>>
>>>> Harold Lim wrote:
>>>>> Hi All,
>>>>>
>>>>>
>>>>> Is it possible in Faban to increase the number
>> of
>>>> concurrent users in the middle of the run? If not,
>> is it
>>>> possible to run 2 Faban instance (maybe on
>> different
>>>> machines) and run the other benchmark run after
>> the first
>>>> one is running a while? Will it cause any problems
>> on
>>>> the faban agents side if 2 instances are
>> talking to
>>>> it?
>>>>> I'm interested to see how the response time
>>>> behaves/pattern when the number of concurrent
>> users change.
>>>>> Thanks,
>>>>> Harold
>>>>>
>>>>>
>>>>>
>>>
>>>
>>
>
>
>
Re: Question about Olio/Faban
Posted by Harold Lim <ro...@yahoo.com>.
I'm curious as to whether the final figure/data will capture the variable workload or just ramp up -> steady state -> ramp down data.
I'm interested in the response time graph of the variable load.
-Harold
--- On Fri, 9/18/09, Akara Sucharitakul <Ak...@Sun.COM> wrote:
> From: Akara Sucharitakul <Ak...@Sun.COM>
> Subject: Re: Question about Olio/Faban
> To: olio-user@incubator.apache.org
> Date: Friday, September 18, 2009, 3:16 PM
> Let me take a look at it. The feature
> needs a lot of work anyway. Will
> take some time, though.
>
> -Akara
>
> Harold Lim wrote:
> > Hi Akara,
> >
> > I did what the doc said but I'm getting a weird
> behavior.
> > My run.xml looks something like this:
> >
> > <fa:runControl>
> > <fa:rampUp>60</fa:rampUp>
> > <fa:variableLoad>true</fa:variableLoad>
> >
> <fa:variableLoadFile>/root/faban/config/profiles/test/variableFile</fa:variableLoadFile>
> > <fa:steadyState>3600</fa:steadyState>
> > <fa:rampDown>60</fa:rampDown>
> > </fa:runControl>
> >
> > My variableFile:
> > 1200, 300
> > 1200, 400
> > 1200, 500
> >
> >
> >
> > The behavior I'm getting is Faban runs the variable
> load first, then after running the variable load, it then
> starts ramping up->steady state -> ramp down.
> >
> > 10:28:25 shirako067
> INFO Variable load
> controller thread started!
> > 10:28:25 shirako067
> INFO Variable load
> controller thread advanced!
> > 10:28:25 shirako067
> INFO Variable load
> controller thread sleeping for 1200 seconds. Setting for 300
> active threads.
> > 10:48:25 shirako067
> INFO Variable load
> controller thread advanced!
> > 10:48:25 shirako067
> INFO Variable load
> controller thread sleeping for 1200 seconds. Setting for 400
> active threads
> > 11:08:25 shirako067
> INFO Variable load
> controller thread advanced!
> > 11:08:25 shirako067
> INFO Variable load
> controller thread sleeping for 1200 seconds. Setting for 500
> active threads.
> > 11:28:25 shirako067
> INFO Started all
> threads; run commences in 2964 ms
> > 11:28:29
> INFO Ramp up started
> > 11:29:28 shirako067
> INFO Ramp up
> completed
> > ...
> >
> >
> >
> > -Harold
> >
> >
> >
> > --- On Thu, 9/3/09, Akara Sucharitakul <Ak...@Sun.COM>
> wrote:
> >
> >> From: Akara Sucharitakul <Ak...@Sun.COM>
> >> Subject: Re: Question about Olio/Faban
> >> To: olio-user@incubator.apache.org
> >> Date: Thursday, September 3, 2009, 12:54 PM
> >> Yes, but the feature is still quite
> >> restrictive. We're building it out better in the
> future.
> >> Please see how to do it here. Let us know if you
> run into
> >> problems.
> >>
> >> http://faban.sunsource.net/1.0/docs/howdoi/loadvariation.html
> >>
> >> Adding yet another Faban instance altogether will
> cause
> >> conflicts.
> >>
> >> -Akara
> >>
> >> Harold Lim wrote:
> >>> Hi All,
> >>>
> >>>
> >>> Is it possible in Faban to increase the number
> of
> >> concurrent users in the middle of the run? If not,
> is it
> >> possible to run 2 Faban instance (maybe on
> different
> >> machines) and run the other benchmark run after
> the first
> >> one is running a while? Will it cause any problems
> on
> >> the faban agents side if 2 instances are
> talking to
> >> it?
> >>> I'm interested to see how the response time
> >> behaves/pattern when the number of concurrent
> users change.
> >>>
> >>> Thanks,
> >>> Harold
> >>>
> >>>
> >>>
> >>
> >
> >
> >
>
>
Re: Question about Olio/Faban
Posted by Akara Sucharitakul <Ak...@Sun.COM>.
Let me take a look at it. The feature needs a lot of work anyway. Will
take some time, though.
-Akara
Harold Lim wrote:
> Hi Akara,
>
> I did what the doc said but I'm getting a weird behavior.
> My run.xml looks something like this:
>
> <fa:runControl>
> <fa:rampUp>60</fa:rampUp>
> <fa:variableLoad>true</fa:variableLoad>
> <fa:variableLoadFile>/root/faban/config/profiles/test/variableFile</fa:variableLoadFile>
> <fa:steadyState>3600</fa:steadyState>
> <fa:rampDown>60</fa:rampDown>
> </fa:runControl>
>
> My variableFile:
> 1200, 300
> 1200, 400
> 1200, 500
>
>
>
> The behavior I'm getting is Faban runs the variable load first, then after running the variable load, it then starts ramping up->steady state -> ramp down.
>
> 10:28:25 shirako067 INFO Variable load controller thread started!
> 10:28:25 shirako067 INFO Variable load controller thread advanced!
> 10:28:25 shirako067 INFO Variable load controller thread sleeping for 1200 seconds. Setting for 300 active threads.
> 10:48:25 shirako067 INFO Variable load controller thread advanced!
> 10:48:25 shirako067 INFO Variable load controller thread sleeping for 1200 seconds. Setting for 400 active threads
> 11:08:25 shirako067 INFO Variable load controller thread advanced!
> 11:08:25 shirako067 INFO Variable load controller thread sleeping for 1200 seconds. Setting for 500 active threads.
> 11:28:25 shirako067 INFO Started all threads; run commences in 2964 ms
> 11:28:29 INFO Ramp up started
> 11:29:28 shirako067 INFO Ramp up completed
> ...
>
>
>
> -Harold
>
>
>
> --- On Thu, 9/3/09, Akara Sucharitakul <Ak...@Sun.COM> wrote:
>
>> From: Akara Sucharitakul <Ak...@Sun.COM>
>> Subject: Re: Question about Olio/Faban
>> To: olio-user@incubator.apache.org
>> Date: Thursday, September 3, 2009, 12:54 PM
>> Yes, but the feature is still quite
>> restrictive. We're building it out better in the future.
>> Please see how to do it here. Let us know if you run into
>> problems.
>>
>> http://faban.sunsource.net/1.0/docs/howdoi/loadvariation.html
>>
>> Adding yet another Faban instance altogether will cause
>> conflicts.
>>
>> -Akara
>>
>> Harold Lim wrote:
>>> Hi All,
>>>
>>>
>>> Is it possible in Faban to increase the number of
>> concurrent users in the middle of the run? If not, is it
>> possible to run 2 Faban instance (maybe on different
>> machines) and run the other benchmark run after the first
>> one is running a while? Will it cause any problems on
>> the faban agents side if 2 instances are talking to
>> it?
>>> I'm interested to see how the response time
>> behaves/pattern when the number of concurrent users change.
>>>
>>> Thanks,
>>> Harold
>>>
>>>
>>>
>>
>
>
>
Re: Question about Olio/Faban
Posted by Harold Lim <ro...@yahoo.com>.
Hi Akara,
I did what the doc said but I'm getting a weird behavior.
My run.xml looks something like this:
<fa:runControl>
<fa:rampUp>60</fa:rampUp>
<fa:variableLoad>true</fa:variableLoad>
<fa:variableLoadFile>/root/faban/config/profiles/test/variableFile</fa:variableLoadFile>
<fa:steadyState>3600</fa:steadyState>
<fa:rampDown>60</fa:rampDown>
</fa:runControl>
My variableFile:
1200, 300
1200, 400
1200, 500
The behavior I'm getting is Faban runs the variable load first, then after running the variable load, it then starts ramping up->steady state -> ramp down.
10:28:25 shirako067 INFO Variable load controller thread started!
10:28:25 shirako067 INFO Variable load controller thread advanced!
10:28:25 shirako067 INFO Variable load controller thread sleeping for 1200 seconds. Setting for 300 active threads.
10:48:25 shirako067 INFO Variable load controller thread advanced!
10:48:25 shirako067 INFO Variable load controller thread sleeping for 1200 seconds. Setting for 400 active threads
11:08:25 shirako067 INFO Variable load controller thread advanced!
11:08:25 shirako067 INFO Variable load controller thread sleeping for 1200 seconds. Setting for 500 active threads.
11:28:25 shirako067 INFO Started all threads; run commences in 2964 ms
11:28:29 INFO Ramp up started
11:29:28 shirako067 INFO Ramp up completed
...
-Harold
--- On Thu, 9/3/09, Akara Sucharitakul <Ak...@Sun.COM> wrote:
> From: Akara Sucharitakul <Ak...@Sun.COM>
> Subject: Re: Question about Olio/Faban
> To: olio-user@incubator.apache.org
> Date: Thursday, September 3, 2009, 12:54 PM
> Yes, but the feature is still quite
> restrictive. We're building it out better in the future.
> Please see how to do it here. Let us know if you run into
> problems.
>
> http://faban.sunsource.net/1.0/docs/howdoi/loadvariation.html
>
> Adding yet another Faban instance altogether will cause
> conflicts.
>
> -Akara
>
> Harold Lim wrote:
> > Hi All,
> >
> >
> > Is it possible in Faban to increase the number of
> concurrent users in the middle of the run? If not, is it
> possible to run 2 Faban instance (maybe on different
> machines) and run the other benchmark run after the first
> one is running a while? Will it cause any problems on
> the faban agents side if 2 instances are talking to
> it?
> >
> > I'm interested to see how the response time
> behaves/pattern when the number of concurrent users change.
> >
> >
> > Thanks,
> > Harold
> >
> >
> >
>
>
Re: Question about Olio/Faban
Posted by Akara Sucharitakul <Ak...@Sun.COM>.
Yes, but the feature is still quite restrictive. We're building it out
better in the future. Please see how to do it here. Let us know if you
run into problems.
http://faban.sunsource.net/1.0/docs/howdoi/loadvariation.html
Adding yet another Faban instance altogether will cause conflicts.
-Akara
Harold Lim wrote:
> Hi All,
>
>
> Is it possible in Faban to increase the number of concurrent users in the middle of the run? If not, is it possible to run 2 Faban instance (maybe on different machines) and run the other benchmark run after the first one is running a while? Will it cause any problems on the faban agents side if 2 instances are talking to it?
>
> I'm interested to see how the response time behaves/pattern when the number of concurrent users change.
>
>
> Thanks,
> Harold
>
>
>