:S you are right...
I only try to register my users, make one or two calls between users
registered and deregistere them...
I'm going to change ratios and test it again.
Thanks
On Mon, Oct 13, 2008 at 3:08 PM, Verbeiren, David <[EMAIL PROTECTED]
> wrote:
> 2008-10-13 10:44:08.255: ERROR: Scenario ratios do not sum up to 100%:
> 200.00000000000000000000.
> 2008-10-13 10:44:08.255: ERROR: Scenario ratios do not sum up to 100%:
> 200.00000000000000000000.
>
> This indicates your scenario ratios are not correct. The Problem is in run
> #1. You specified ims_uac for 50% and ims_dereg for 50%. However, ratios
> from previous run are preserved if not specified, hence you also inherit
> ims_reg for 100% from run #0. This gives a total of more than 100% which is
> not valid.
>
> So if you really want 50% dereg, 50% uac, you should force ims_reg to 0%.
>
> However, such a mix will obviously not work since you would be leaking
> users very quickly by deregistering them without ever re-registering them
> afterwards. So it's probably not what you intended to do anyway.
>
> Also, I don't think you ever said whether the calls (even if only 1 or 2)
> were actually successful or not.
>
> -David
>
> ________________________________________
> From: Vanessa Tejada Muñoz [mailto:[EMAIL PROTECTED]
> Sent: lundi 13 octobre 2008 13:54
> To: Verbeiren, David
> Cc: sipp-users@lists.sourceforge.net
> Subject: Re: [Sipp-users] Run basic scenarios IMS bench.
>
> Hi again,
>
> I would like to know which is the minimun value of users to work with the
> basic scenarios in IMS bench. I attach my manager.xml. I made a lot of
> chages in the parameters but maybe they are wrong now... I can not make it
> work... Now, I have 8 users.
>
> My common error_log is:
>
> 2008-10-13 10:43:59.953: Created CConsole 0x81f5570.
> 2008-10-13 10:43:59.954: Set TSID: slot:0 TS1 (1).
> 2008-10-13 10:43:59.954: accept return 20 [IP4: 192.168.34.36:53264].
> 2008-10-13 10:43:59.954: ~ASSIGNID=0xffffffff - 0x1.
> 2008-10-13 10:43:59.954: ~ASSIGNID=0x1 - 0xffffffff.
> 2008-10-13 10:43:59.954: Set TSID: slot:1 TS1 (0).
> 2008-10-13 10:44:08.255: ERROR: Scenario ratios do not sum up to 100%:
> 200.00000000000000000000.
> 2008-10-13 10:44:08.255: ERROR: Scenario ratios do not sum up to 100%:
> 200.00000000000000000000.
> 2008-10-13 10:44:09.248: User pool id[2] is empty -> Quitting!.
> 2008-10-13 10:44:09.248: Quitting!.
> 2008-10-13 10:44:19.805: sipp.cpp: EXIT_TEST_RES_UNKNOWN quitting=1.
> 2008-10-13 10:44:19.805: final cleanup.
>
>
> Thanks in advace
>
> On Thu, Oct 9, 2008 at 2:29 PM, Verbeiren, David <
> [EMAIL PROTECTED]> wrote:
> Well... AFAIK the vanilla SIPp only supports one scenario at a time per
> SIPp process instance. Hence you have 2 options:
> - First run a registration scenario, then a calling or messaging
> one. It won't take care of avoiding that the same user participates in
> multiple calls or any such kind of coordination though. Also you would run
> into problem in case you need data from the execution of the reg scenario to
> be used in the subsequent ones. (like storing the Service-Route header as is
> done in IMS Bench SIPp - but if you know you always get the same value from
> your IMS core, there isn't much point in storing it in a dynamic
> variable...)
> - Create a giant scenario that combines all the scenarios you need
> and uses branching to sequence them. I know folks designing the SPEC SIP
> benchmark are using this technique.
>
> IMS Bench SIPp is better at emulating actual users but you probably can get
> around limitations of the vanilla SIPp.
>
> I your intention is to run benchmarks and not just simple test, sticking to
> IMS Bench SIPp is a good thing. But if you just want to test some scenarios,
> vanilla SIPp seems to simplest way to me.
>
> -David
>
> ________________________________________
> From: Vanessa Tejada Muñoz [mailto:[EMAIL PROTECTED]
> Sent: jeudi 9 octobre 2008 13:55
>
> To: Verbeiren, David
> Subject: Re: [Sipp-users] Run basic scenarios IMS bench.
>
> Perfect,
>
> I think the same... Then you think that I can make the same test with SIPp
> (not IMS Bench) for this amount of users, and then, If I have more, trying
> it with IMS Bench.
> The labels in xml of sipp are different from ims_bench but I can do the
> same actions more or less, like for example use variables for all scenarios,
> and these king of things, Is it?
>
> Thanks for all David.
> On Thu, Oct 9, 2008 at 1:19 PM, Verbeiren, David <
> [EMAIL PROTECTED]> wrote:
> The reason is that the check for empty pool is done at time of selecting a
> user and actually checks for the pool to be empty after picking the user.
> With 2 users, and trying to register both, you will inevitably empty the
> pool and this is why you get the failure (look in user.cpp:pick_user()
> function).
>
> To get over this error, you'd need to define at least 3 users...
>
> This is all due to the fact that the normal operation uses statistical
> selection of scenario, user, instantaneous rate and hence must be
> dimensioned, in terms of users, well over the expected number of users to
> operate correctly. Hence checking for a pool becoming empty is a valid check
> under such assumption. Again, your current usage just doesn't match the
> design objectives of the tool.
>
> -David
>
>
> ________________________________________
> From: Vanessa Tejada Muñoz [mailto:[EMAIL PROTECTED]
> Sent: jeudi 9 octobre 2008 12:50
>
> To: Verbeiren, David
> Subject: Re: [Sipp-users] Run basic scenarios IMS bench.
>
> I know It doesn't really sense, but it is the first step to add more users
> in the scenarios when we created more Service Profiles with more users in
> the IMS Core. The goal is, if you can do it with 2 users like a simple call
> between them, you will can do it for much more.
>
> I'm sorry so much, but I can't achiev it. I attach manager.xml, manager.log
> and sipp errors log. I changed the values such as you said.
>
> Thanks again.
> On Thu, Oct 9, 2008 at 11:16 AM, Verbeiren, David <
> [EMAIL PROTECTED]> wrote:
> If you only have 2 users in pool 0 and you run registration scenarios at 1
> scenario attempt per second, your registration phase may only last for 2
> seconds because after that you pool of users available to perform
> registration scenarios is empty (both users are registered by then). If you
> configure it so, you would not run out of users during the registration
> phase. If you then want to have a subsequent phase with calling scenarios,
> assuming again 1 call per second, you'd have to make sure the call duration
> is always below 1 second because 1 second after the first call makes the 2
> users busy (by moving them to a specific pool for users who are busy on a
> call), you'd need to have them free again to start the next call. This also
> means you should use constant call duration instead of statistical one.
>
> Maybe this is possible but, to be honest, trying to use IMS Bench SIPp with
> just 2 users really doesn't make much sense. Again, why are you trying with
> IMS Bench for this type of exercise when the "vanilla" SIPp (no manager, no
> user pools, just a simple input file with the data for your 2 users)?
>
> Is there something specific in IMS Bench SIPp that you need and that is not
> there in the vanilla SIPp?
>
> -David
>
> P.S. By "vanilla SIPp", I mean the main SIPp code branch.
>
> ________________________________________
> From: Vanessa Tejada Muñoz [mailto:[EMAIL PROTECTED]
> Sent: jeudi 9 octobre 2008 11:03
>
> To: Verbeiren, David
> Cc: sipp-users@lists.sourceforge.net
> Subject: Re: [Sipp-users] Run basic scenarios IMS bench.
>
> Hi David,
>
> I would like to know why I can test the basic scenarios with only 2 users.
> I think that it would be a good idea create a kind of label to control the
> number of users. I have always the same error because my pool is empty... I
> found another question in the mailing list about this issue. I don't
> understant why I can't test the scenarios with a limit number of users. Why
> is mandatory have thousands of users? It may people want to check its IMS
> core with a bit users before.
>
> I need to know if It is imposible run basic scenarios for only 2 users. I
> prefer know it instead of change the manager.xml for nothing.
>
>
> Thanks a lot for your assistance.
> On Wed, Oct 8, 2008 at 12:50 PM, Vanessa Tejada Muñoz <[EMAIL PROTECTED]>
> wrote:
> But the thing is, you should can scecific the number of calls you want like
> in Sipp, because I only want to test a call between two users throught IMS
> Core. I understand that the ideal test requires a huge number of users, but
> that is the requirement of my test now :S
>
> On Wed, Oct 8, 2008 at 12:41 PM, Verbeiren, David <
> [EMAIL PROTECTED]> wrote:
> I don't understand what your goal is. IMS Bench SIPp is meant to provide a
> benchmark test system. It will try to load the system under test with many
> scenarios and to this end, it requires many users. If you don't intend to
> benchmark, then you'd be better off with the vanilla SIPp, IMHO. I think
> there are "IMS enabled" scenarios available for vanilla SIPp as well. You
> may want to look on the open-ims core wiki or the SIPp wiki.
>
>
> ________________________________________
> From: Vanessa Tejada Muñoz [mailto:[EMAIL PROTECTED]
> Sent: mercredi 8 octobre 2008 12:35
>
> To: Verbeiren, David
> Subject: Re: [Sipp-users] Run basic scenarios IMS bench.
>
> Yes, I know, but in my company, they only want to check a call between 2
> users and compare the SIPp test with Softphones. They have created 2 users
> in the IMS Core that are associated with a Service Profile and Application
> Server. That is why I have only two users... And I'm trying to down rates in
> my manager.xml... I don't know it it will work or we will need mandatory
> more users.
> On Wed, Oct 8, 2008 at 12:29 PM, Verbeiren, David <
> [EMAIL PROTECTED]> wrote:
> 2 users ???? Why would benchmark with 2 users? You need thousands of users.
> If you use the provided ims_bench.pl script (Perl) to configure the
> benchmark, it will create the users file for you. Alternatively, I think the
> source tree includes a tiny user_gen.pl script that can also be used (as-is
> or as basis) to generate the users inf file.
>
> -David
>
> ________________________________________
> From: Vanessa Tejada Muñoz [mailto:[EMAIL PROTECTED]
> Sent: mercredi 8 octobre 2008 12:20
> To: Verbeiren, David
> Cc: sipp-users@lists.sourceforge.net
>
> Subject: Re: [Sipp-users] Run basic scenarios IMS bench.
>
> The problem is in rates. I have only 2 users. When a I changed the cps of
> register scenario in manager, the manager execution continued a bit more.
> I'll work in this issue. I hope do not disturb you again.
>
> Thanks David.
>
> Regards,
> On Wed, Oct 8, 2008 at 10:57 AM, Verbeiren, David <
> [EMAIL PROTECTED]> wrote:
> Hi Vanessa,
>
> You have to understand the key concept of user pools in IMS Bench SIPp. The
> whole benchmark is very "user centric" with the test system acting as a user
> population. In order to be able to correctly emulate user behaviours, the
> system must keep track of the user state in some way (so as to, for example,
> place calls to registered or non registered users in the proportion
> specified by the scenario execution ratios; or to only deregister users that
> were previously registered). It is actually the scenarios that have to
> maintain this user state because only the scenarios know what they intend to
> do (SIPp has no knowledge of "being registered" or anything like that so it
> remains fully flexible). The way IMS Bench SIPp provides support for this to
> the scenario designer is by means of those user pools. You can define
> yourself what state (or possibly combination of states) a pool represent. In
> standard scenarios, pool 0 is the pool of non-registered users. Hence the
> ims_reg scenario takes users from this pool as you can see in the following
> scenario fragment (from beginning of ims_reg scenario):
> <nop>
> <action>
> <assign_user pool="0" scheme="rand_uni"/> <!-- 'Unregistered' user
> -->
> <move_user pool="1"/> <!-- Registration ongoing
> -->
> </action>
> </nop>
>
> It also puts the user into a temporary pool, pool 1, in order to avoid
> starting two registrations in parallel for the same user. Because an
> instance of the ims_reg scenario moves the user into pool 1 as soon as it
> has selected him to perform a registration, the same user cannot be picked
> again by another instance of ims_reg scenario because the scenario always
> takes its user from pool 0, not pool 1.
>
> I hope I'm making sense here. This is also explained in the IMS Bench SIPp
> online documentation (
> http://sipp.sourceforge.net/ims_bench/reference.html#User+oriented+mode)
> but I understand there is a lot of information to digest there.
>
> So the error
> 2008-10-07 14:27:33.624: User pool id[0] is empty -> Quitting!.
> means that the ims_reg scenario that was trying to run could not find any
> user in pool 0. Without a user, it can't run because its purpose is to
> register the user. If this happens right away when you start your test, then
> it probably means the file you specify by means of the -user_inf SIPp
> command line attribute is not correct. In this file, the first column
> indicates the initial pool in which the user is put. If you start by a
> registration phase (as in the standard benchmark), then all users should
> initially live in pool 0.
>
> If it happens after some time of correct execution, then it probably means
> you did not provide enough users in order to run the registration phase at
> the specified rate and for the specified duration. (e.g. 100 registrations/s
> for a 120 s registration phase requires at least 12000 users in the
> -user_inf specified file; and if you keep doing registrations together with
> other scenarios after the initial registration phase, then you should have
> slightly more users in order to avoid running out of users in pool 0 in the
> subsequent phases).
>
> -David
>
> ________________________________________
> From: Vanessa Tejada Muñoz [mailto:[EMAIL PROTECTED]
> Sent: mardi 7 octobre 2008 14:23
> To: Verbeiren, David
> Subject: Re: [Sipp-users] Run basic scenarios IMS bench.
>
> David,
>
> sorry for my constant questions :S I found an error in usuarios.csv. I
> create users.inf as I read in documentation, now I'm registered with the
> proxy but the execution don't continue with INVITE message. The error log
> said:
>
>
> Epi:/home/ims1/sipp_3_1_ims_bench/ims_bench # cat 20794_errors.log
> sipp: The following events occured:
> 2008-10-07 14:27:31.124: Created CConsole 0x81f5580.
> 2008-10-07 14:27:31.125: Set TSID: slot:0 TS1 (1).
> 2008-10-07 14:27:31.125: accept return 18 [IP4: 192.168.34.36:43445].
> 2008-10-07 14:27:31.125: ~ASSIGNID=0xffffffff - 0x1.
> 2008-10-07 14:27:31.125: ~ASSIGNID=0x1 - 0xffffffff.
> 2008-10-07 14:27:31.125: Set TSID: slot:1 TS1 (0).
> 2008-10-07 14:27:33.624: User pool id[0] is empty -> Quitting!.
> 2008-10-07 14:27:33.625: Quitting!.
> 2008-10-07 14:27:33.657: sipp.cpp: EXIT_TEST_RES_UNKNOWN quitting=1.
> 2008-10-07 14:27:33.658: final cleanup.
>
> My document ims_reg is equal to the IMS bench ims_reg.xml, I add the line
> "[authentication [EMAIL PROTECTED] password=[field4]
> challenge_from="u2" challenge_type=401]" for the none, but I don't
> understand what is the message "2008-10-07 14:27:33.624: User pool id[0] is
> empty -> Quitting!."
>
> Thanks
> On Tue, Oct 7, 2008 at 1:13 PM, Vanessa Tejada Muñoz <[EMAIL PROTECTED]>
> wrote:
> Sorry, I forgot attach again ims_reg.xml
>
> On Tue, Oct 7, 2008 at 1:06 PM, Vanessa Tejada Muñoz <[EMAIL PROTECTED]>
> wrote:
> As I read in the documentation, I execute manager.xml then sipp comand and
> finally press 'e' key in manager execution and finally the sipp comand, but
> the manager stop at time and I have to finish the test with crtl+c because
> nothing change.
>
> I attach the screens, usuarios.csv, ims_reg.xml (because I changed one line
> in REGISTER authentication) and log_sipp.
>
> On Tue, Oct 7, 2008 at 12:36 PM, Verbeiren, David <
> [EMAIL PROTECTED]> wrote:
> Could you please send screen output and log files from manager and sipp ?
>
> Note that, as described in
> http://sipp.sourceforge.net/ims_bench/reference.html#Running, once all
> desired components are started (and you see they all connected to the
> manager) you should press the 'e' key in the manager console.
> (alternatively, you can configured the expected number of SIPp instances and
> have the manager start automatically once they all have connected).
>
> -David
>
> ________________________________________
> From: Vanessa Tejada Muñoz [mailto:[EMAIL PROTECTED]
> Sent: mardi 7 octobre 2008 12:16
> To: sipp-users@lists.sourceforge.net
> Subject: [Sipp-users] Run basic scenarios IMS bench.
>
> Hi all,
>
> I'm trying to run the basic scenarios in IMS Bench which appears in a
> folder called scen when you install it all.
>
> I execute ./manager - f manager.xml and then the comand: ./sipp
> ip_core_ims:4060 -rmctrl ip_manager_machine:5000 -user_inf scen/usuarios.csv
> -i ip_manager_machine -p 5000
>
> But the problem is that anything happen... Can anybody tell me how to test
> the basic scripts please? I think that understanding the basic funcionality
> I can create my own scenarios
>
> Best regards
>
>
> --
>
> Vanessa Tejada
> ---------------------------------------------------------------------
> Intel Corporation NV/SA
> Rond point Schuman 6, B-1040 Brussels
> RPM (Bruxelles) 0415.497.718.
> Citibank, Brussels, account 570/1031255/09
>
> This e-mail and any attachments may contain confidential material for
> the sole use of the intended recipient(s). Any review or distribution
> by others is strictly prohibited. If you are not the intended
> recipient, please contact the sender and delete all copies.
>
>
--
Vanessa Tejada
-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
Sipp-users mailing list
Sipp-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sipp-users
