AW: Re: [AGENT++] Help with getting agent++ work

[fock at agentpp.com]

Hi Subbha,

The threading works on windows and and Linux of course.
However you need to use the appropriate settings for
both platforms as they are defined by default.

Best regards,
Frank

>
>Hello Frank! Thanks for your tip- I figured that the problems I was facing 
>were because of the fact that snmp++ and agent++ were not being built with 
>the same settings. Once I fixed that, I was able to get the atm_mib example 
>and another very simple example that I wrote working. But this would work 
>only with the _NO_THREADS defined. Do you know if the thread pooling works 
>on windows and linux? Without the _NO_THREADS defined, both the atm_mib 
>example and my simple example will crash when they try to handle a incoming 
>request by calling threadpool->execute. 
>
>Thanks for your help!
>Subha
>
>On 7/27/05, Frank Fock  wrote:
>> 
>> Hi Subbha,
>> 
>> Since you wrote that you had the problem with any example I thought
>> it that the problem is related to the #defines. Knowing that you tried
>> the cmd_exe_mib example it may also be a bug in that example (it
>> is very old and probably not well maintained). Please try the atm
>> example first. I will have a look at the cmd_mib example and either
>> fix it or remove it for the next release.
>> 
>> Best regards,
>> Frank
>> 
>> Subha Velayutham wrote:
>> 
>> >Hi Frank! I tried doing what you suggested - downloaded snmp++ and 
>> agent++
>> >again and built them and tried running the cmd_exe_mib example - when 
>> trying
>> >to handle a incoming request , the program was crashing in
>> >mib.cpp:process_request at MibTask* mt = new MibTask(call);
>> >threadPool->execute(mt);
>> >
>> >Then I tried adding _NO_THREADS to the Preprocessor definitions of SNMP++
>> >project and rebuilt SNMP++ first followed by agent++ and then cmd_exe_mib 
>> -
>> >this time around the usm variable is null within UsmUserTable constructor 
>> in
>> >v3_mib.cpp and therefore the program crashes even before it gets to the 
>> part
>> >where it has to listen for requests.
>> >
>> >Basically I am getting the same behavior as before.
>> >
>> >What am I doing wrong?
>> >
>> >Thanks,
>> >Subha
>> >On 7/26/05, Frank Fock  wrote:
>> >
>> >
>> >>Hi,
>> >>
>> >>Please download SNMP++ and AGENT++ again and compile it
>> >>ina different place. It seems that you have changed some #defines
>> >>without recompiling everything and thus using a consistent set of
>> >>#defines.
>> >>
>> >>Best regards,
>> >>Frank
>> >>
>> >>Subha Velayutham wrote:
>> >>
>> >>
>> >>
>> >>>I am using snmp++ version 3.2.17 and agent++ version 3.5.23.a on
>> >>>windows/linux and running into a lot of issues even getting the 
>> examples
>> >>>that come with the agent++ source work.
>> >>>
>> >>>I am using the atm example from the agent++\examples folder. The same
>> >>>behavior happens with the other examples too. I really need help.
>> >>>
>> >>>If I use SNMPV3, then I get a access violation at line 434(int users =
>> >>>usm->get_user_count();) of V3_mib.cpp because the usm got at line 397 
>> is
>> >>>null. (usm = v3mp->get_usm();) In fact the v3MP created in the main
>> >>>
>> >>>
>> >>routine
>> >>
>> >>
>> >>>of the atm example itself is null.(line 363)
>> >>>
>> >>>If I define _NO_SNMPV3 and if I dont define _NO_THREADS I get an access
>> >>>violation exception in mib.cpp at line 3375.
>> >>>3374: MibTask* mt = new MibTask(call);
>> >>>3375: threadPool->execute(mt);
>> >>>
>> >>>If I define _NO_SNMPV3 and if I do define _NO_THREADS then the
>> >>>Snmpx::receive(struct timeval *tvptr, Pdux& pdu, UdpAddress& fromaddr,
>> >>>snmp_version& version, OctetStr& community) method returns with
>> >>>SNMP_CLASS_TL_FAILED from line 1404.
>> >>>
>> >>>1399: nfound = select(max_fd+1, &readfds, 0, 0, tvptr);
>> >>>if (nfound == -1)
>> >>>{
>> >>>if (errno != EINTR)
>> >>>1404: return SNMP_CLASS_TL_FAILED;
>> >>>continue;
>> >>>}
>> >>>
>> >>>Can someone help me here? I would really appreciate any help - if you
>> >>>
>> >>>
>> >>have
>> >>
>> >>
>> >>>working examples that would be even better.
>> >>>
>> >>>Thanks,
>> >>>Subha
>> >>>_______________________________________________
>> >>>AGENTPP mailing list
>> >>>AGENTPP at agentpp.org
>> >>>http://lists.agentpp.org/mailman/listinfo/agentpp
>> >>>
>> >>>
>> >>>
>> >>>
>> >>>
>> >>>
>> >>
>> >>
>> >>
>> >_______________________________________________
>> >AGENTPP mailing list
>> >AGENTPP at agentpp.org
>> >http://lists.agentpp.org/mailman/listinfo/agentpp
>> >
>> >
>> >
>> >
>> 
>> 
>>