[SNMP4J] Fw: Re: Thread count leads to SNMP response timeouts in snmp4j version 1.7.1

[loaner45244 at mypacks.net]

Hi Frank,

How can we determine that it is the threading model that may be causing the issues here?
Is there a way to determine what the peak rate is for set/get s and snmp informs that may be handled by the snmp4j stack; if you could point me to a test harness or even a back-off client, that would be great.

-----Original Message-----
>From: Frank Fock <fock at agentpp.com>
>Sent: Jan 28, 2011 2:13 PM
>To: snmp4j at agentpp.org
>Subject: Re: [SNMP4J] Fw: Re: Thread count leads to SNMP response timeouts in snmp4j version 1.7.1
>
>Hi,
>
>It seems that the problem is not caused by the
>SNMP4J API, but in the application.
>
>Best regards,
>Frank
>
>
>On 28.01.2011 21:51, loaner45244 at mypacks.net wrote:
>>
>> -----Forwarded Message-----
>>> Subject: Re: [SNMP4J] Thread count leads to SNMP response timeouts in snmp4j version 1.7.1
>>>
>>> These are our internal threads (ThreadPool) that spawn off and
>>> manage snmp traffic to approximately 1000 peers. The problem is
>>> that as we add additional threads, the org.snmp4j.Snmp object's
>>> ResponseListener callback begins to timeout.
>>>
>>> I did not follow the memory question - the machine has several
>>> Gigs of memory so that should not be an issue but is there any
>>> particular thread based (minimum) limit that I must check?
>>>
>>> Thanks,
>>>
>>> HA
>>>
>>>
>>> -----Original Message-----
>>>> From: Frank Fock<fock at agentpp.com>
>>>> Sent: Jan 25, 2011 5:08 PM
>>>> To: snmp4j at agentpp.org
>>>> Subject: Re: [SNMP4J] Thread count leads to SNMP response timeouts in snmp4j version 1.7.1
>>>>
>>>> Hi,
>>>>
>>>> What do you mean with "thread count"?
>>>> Which class and method do you use to "increase"
>>>> it?
>>>> Why do you increase it?
>>>> (You can process responses of many devices
>>>> with a single thread)
>>>> Do you have enough memory for the threads?
>>>> (150*1MB = 150MB)
>>>>
>>>> Best regards,
>>>> Frank
>>>>
>>>>
>>>> On 26.01.2011 00:28, loaner45244 at mypacks.net wrote:
>>>>> Hi All,
>>>>>
>>>>> We're running an old version of snmp4j (v1.7.1) where we are
>>>>> see-ing a peculiar behavior with respect to how snmp4j handles
>>>>> SNMP responses if the thread count is increased.
>>>>>
>>>>> Our default setting is 50 threads and when that value is increased to 150,
>>>>> the SNMP response timeouts expire. We know that the peer has the ability
>>>>> to send back responses even with 150 threads so any help in debugging this
>>>>> would be appreciated.
>>>>>
>>>>> In particular:
>>>>>
>>>>> 1. The usual suspect is that this may be just inherent behavior
>>>>>      of the old stack. Can anyone confirm if they have seen similar issues?
>>>>>
>>>>> 2. What kind of debugging can be employ to understand this behavior? Are there
>>>>>      logs that could identify the reason why timeouts may be occurring?
>>>>>
>>>>> 3. Are there any known enhancements to the latest snmp4j code that address
>>>>>      performance and timeout issues seen in the past? (to make it worth pursuing).
>>>>>
>>>>> Thanks,
>>>>>
>>>>> HA
>>>>> _______________________________________________
>>>>> SNMP4J mailing list
>>>>> SNMP4J at agentpp.org
>>>>> http://lists.agentpp.org/mailman/listinfo/snmp4j
>>>>
>>>> --
>>>> AGENT++
>>>> http://www.agentpp.com
>>>> http://www.snmp4j.com
>>>> http://www.mibexplorer.com
>>>> http://www.mibdesigner.com
>>>>
>>>> _______________________________________________
>>>> SNMP4J mailing list
>>>> SNMP4J at agentpp.org
>>>> http://lists.agentpp.org/mailman/listinfo/snmp4j
>>
>> _______________________________________________
>> SNMP4J mailing list
>> SNMP4J at agentpp.org
>> http://lists.agentpp.org/mailman/listinfo/snmp4j
>
>-- 
>AGENT++
>http://www.agentpp.com
>http://www.snmp4j.com
>http://www.mibexplorer.com
>http://www.mibdesigner.com
>
>_______________________________________________
>SNMP4J mailing list
>SNMP4J at agentpp.org
>http://lists.agentpp.org/mailman/listinfo/snmp4j