[SNMP4J] net-snmp and snmp4j agentx
Rob Morton
rob.d.morton at alcatel-lucent.com
Thu Jun 7 15:44:07 CEST 2012
Hi Frank,
I will upgrade to net-snmp 5.5 and try it with that version.
Thank you :-)
Regards,
Rob
On 6/6/2012 5:51 PM, Frank Fock wrote:
> Hi Rob,
>
> Bad news, it seems that this is simply a memory allocation bug in
> NET-SNMP. NET-SNMP 5.0.9 is very old and there are a lot of fixes
> regarding AgentX in newer versions. Are you able to upgrade?
>
> Best regards,
> Frank
>
> Am 06.06.2012 19:43, schrieb Rob Morton:
>> Hi Frank,
>>
>> First, thank you very much for looking into this issue. It is very
>> much appreciated.
>>
>> Second, I have attached a master agent log(I use log4j too and had
>> to connect in your logsnmp4jfactory)...it shows the agent starting up
>> and the log messages generated as a result of the net-snmp starting
>> up and connecting then registering its MIB regions...later in the
>> log...it then reflects DEBUG info when I send in the snmpget for the
>> ifNumber OID as stated earlier in this email thread.
>>
>> I removed my server's log entries as best I could but there might
>> still be a few entries reflecting my code...it will be prefixed with
>> com.alcatel.tpapps.*
>>
>> Thanks,
>> Rob
>>
>> On 6/6/2012 11:55 AM, Frank Fock wrote:
>>> Hi Rob,
>>>
>>> The log of the sub-agent shows, that there is probably a region
>>> issue. To get the root cause, the log of the master agent is very
>>> important.
>>>
>>> It is likely, that we can find a workaround for that issue.
>>> NET-SNMP cannot handle big sub OIDs (between 2^31
>>> and 2^32-1) in registrations/requests correctly in all
>>> situations.
>>>
>>> Best regards,
>>> Frank
>>>
>>> Am 06.06.2012 15:47, schrieb Rob Morton:
>>>> Hi Frank,
>>>>
>>>> Thank you for responding...
>>>>
>>>> Here are the details of the issue I'm facing:
>>>>
>>>> * I am using net-snmp v5.0.9 on Solaris 10 configured as a
>>>> sub-agent to my AgentXMasterAgent-based master agent via the
>>>> command:
>>>>
>>>> /usr/sfw/sbin/snmpd -X -DALL
>>>>
>>>> * with /etc/sma/snmp/snmpd.conf having the following entries:
>>>>
>>>> #to avoid re-registrations when I suspend the master agent's
>>>> threads in IntelliJ's debugger
>>>> agentxPingInterval 60000
>>>> agentXSocket tcp:localhost:1705
>>>>
>>>> * I have no issues with connection nor registration...I have
>>>> attached a SNMP walk of the master in which I grep'd for agentx
>>>> to observe what was registered by the sub-agent (see attached
>>>> snmpwalk-grepedForAgentX.log)...I used the following command:
>>>>
>>>> /usr/sfw/bin/snmpwalk -v2c -c<community string>-M
>>>> /etc/sma/snmp/mibs -m ALL <host IP@> 1 | grep agentx >
>>>> /opt/snmpwalk-grepedForAgentX.log
>>>>
>>>> * now.../_the problem_/...It is as follows...I am sending in a
>>>> simple snmp get request to my master snmp agent (for an OID
>>>> managed by the net-snmp sub-agent) with the following command:
>>>>
>>>> /usr/sfw/bin/snmpget -v2c -c<community string> -r 0 -t 60 <host
>>>> IP@> 1.3.6.1.2.1.2.1.0 <-- this OID is IF-MIB::ifNumber.0
>>>>
>>>> * to which I receive the following response:
>>>>
>>>> IF-MIB::ifNumber.0 = No Such Object available on this agent at this
>>>> OID <--THE PROBLEM
>>>>
>>>> * I have attached the net-snmp debug log (see attached snmpd.log)
>>>> for the above SNMP get request.
>>>>
>>>> * if I send in a simple get for an OID managed by the master
>>>> locally, I have no issue:
>>>>
>>>> /usr/sfw/bin/snmpget -r 0 -t 600 -v2c -c<community string><host
>>>> IP@>sysContact.0
>>>> SNMPv2-MIB::sysContact.0 = STRING: bob
>>>>
>>>> * I don't currently have a log for the master agent (I have
>>>> issues linking in the log4j with my code base...am working on
>>>> it) _but_ I have stepped through the code extensively and
>>>> observed no exceptions or strange code paths for this
>>>> request...it simply assesses that it must send a GET PDU to the
>>>> sub-agent and does just that...I have also verified each byte
>>>> in the encoded PDU...it all looks good.
>>>>
>>>> If more info is required please don't hesitate to ask.
>>>>
>>>> Thank you,
>>>> Rob
>>>>
>>>> On 6/6/2012 5:39 AM, Frank Fock wrote:
>>>>> Hi Rob,
>>>>>
>>>>> With the NET-SNMP AgentX master and sub-agents several
>>>>> issues are known. Any of those, are caused by NET-SNMP's
>>>>> incorrect registration algorithms or simply other bugs
>>>>> which are not fixed for years now.
>>>>>
>>>>> For several issues there exists workarounds. So if you are
>>>>> using NET-SNMP as a subagent to SNMP4J-AgentX, nearly
>>>>> everythin should work. You can search for "NET-SNMP"
>>>>> in the sources of SNMP4J-AgentX to locate the various
>>>>> workarounds.
>>>>>
>>>>> We could probably better help, if you provide more details
>>>>> on the problem (master log output of the failed request
>>>>> as well as subagent output, also who is master and who
>>>>> subagent, etc.)
>>>>>
>>>>> Best regards,
>>>>> Frank
>>>>>
>>>>> Am 06.06.2012 02:03, schrieb Rob Morton:
>>>>>> Hi,
>>>>>>
>>>>>> What version(s) of net-snmp are supported by snmp4j's agentxv2.1.0?
>>>>>>
>>>>>> I ask because I am getting peculiar 'get' failures on Solaris's
>>>>>> net-snmpv5.0.9. I get these errors after successful connection,
>>>>>> registration, pings, etc...I perform an snmpget on IF-MIB::ifNumber.0
>>>>>> (just an example) and get "IF-MIB::ifNumber.0 = No Such Object available
>>>>>> on this agent at this OID" This OID region is indeed registered.
>>>>>>
>>>>>> Thanks,
>>>>>> Rob
>>>>>>
>>>>>> _______________________________________________
>>>>>> SNMP4J mailing list
>>>>>> SNMP4J at agentpp.org
>>>>>> http://lists.agentpp.org/mailman/listinfo/snmp4j
>>>>
>>>
>>> --
>>> ---
>>> AGENT++
>>> Maximilian-Kolbe-Str. 10
>>> 73257 Koengen, Germany
>>> https://agentpp.com
>>> Phone: +49 7024 8688230
>>> Fax: +49 7024 8688231
>>
>
> --
> ---
> AGENT++
> Maximilian-Kolbe-Str. 10
> 73257 Koengen, Germany
> https://agentpp.com
> Phone: +49 7024 8688230
> Fax: +49 7024 8688231
More information about the SNMP4J
mailing list