[SNMP4J] Filling a shared table with two index columns from several processes not working
HAAS Christian
Christian.Haas at frequentis.com
Mon Oct 1 12:03:48 CEST 2012
Hello again!
Months later I've been assigned again to this topic - and still struggle with the solution. (I've included the complete conversation below.)
For a start, I've switched the dependency of snmp4j from 1.10.2 to 2.1.0, which includes your extension code.
First I was struggling with getting the new IndexStrategy.anyNonAllocatedSubIndex into the DefaultAgentXSharedMOTable. Because at construction time, this one doesn't have a AgentXSharedMOTableSupport object set - and setting one doesn't seem to be helpful either, because AgentXSubagent.registerRegions() creates a new object and directly sets it - right before the call to AgentXSubagent.registerSharedTableRows().
Since I found no way to access or modify the table support, I created an extension of DefaultAgentXSharedMOTable which overwrites setAgentXSharedMOTableSupport() and sets the IndexStrategy on the passed object before passing it on to super.
I don't think that this is the proper way, but couldn't see any other. At least this then worked for the first process to let it fill in its first rows with combined keys.
But starting the second process that wanted to extend this table further failed again; The response code in AgentXSharedMOTableSupport.allocateIndex() was 260. Is this the problem you mentioned that NetSNMP itself can't work with combined keys?
As a cross test, I left the first process running and switched the IndexStrategy to IndexStrategy.anyNonAllocatedSubIndex . This one would be accepted by NetSNMP, but causes then an InvalidArgumentException in MOTableIndex.getIndexOID(), called from AgentXSharedMOTableSupport.allocateIndex(), while handling the response. The reason here is that the index has two entries, but allocateIndex() had the vbs variable reset to be an array of size 1 just before (in the switch for the strategy).
(This exception was also thrown when I was using all the default code, i.e. no overriding of setAgentXSharedMOTableSupport() )
I'm sorry, but I'm still lost about how to use this. Do you have any advice what I should change?
Kind regards,
ch
____________________________________________________
Christian Haas
Software Engineer
FREQUENTIS AG
Innovationsstraße 1, 1100 Vienna, Austria
Phone +43-1-811 50 – 8353
Mobile +43-664-60 850 – 8353
Fax +43-1-811 50 – 77 8353
Web www.frequentis.com
E-Mail christian.haas at frequentis.com
Handelsgericht Wien (Vienna Commercial Court): FN 72115 b
DVR 0364797, ATU 14715600
____________________________________________________
Diese E-Mail könnte vertrauliche und/oder rechtlich geschützte Informationen enthalten. Wenn Sie nicht der richtige Adressat sind oder diese E-Mail irrtümlich erhalten haben, informieren Sie bitte sofort den Absender und vernichten Sie diese E-Mail. Das unerlaubte Kopieren sowie die unbefugte Weitergabe dieser E-Mail sind nicht gestattet.
This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorised copying, disclosure or distribution of the material in this e-mail is strictly forbidden.
-----Original Message-----
From: snmp4j-bounces at agentpp.org [mailto:snmp4j-bounces at agentpp.org] On Behalf Of Frank Fock
Sent: Montag, 05. März 2012 17:23
To: snmp4j at agentpp.org
Subject: Re: [SNMP4J] Filling a shared table with two index columns from several processes not working
Hi,
will post a link to the SNAPSHOT sources here once,
I have finished the implementation.
I am currently extending the test subagent to be able
to test and illustrate the behaviors.
Mostr likely, the AgentXSharedTableSupport class
will be extended to support four index allocation
strategies. The default one, will be suitable for
most use cases then (including master/child table).
Thus, API users will not have to change their code
in most cases.
Best regards,
Frank
Am 05.03.2012 14:26, schrieb HAAS Christian:
> Hello there!
>
> From: snmp4j-bounces at agentpp.org [mailto:snmp4j-bounces at agentpp.org] On Behalf Of Frank Fock
>> You are facing the same issue as described in http://lists.agentpp.org/pipermail/snmp4j/2012-March/004789.html
>>
>> I understand that this situation is not covered userfriendly by SNMP4J-AgentX as implementing a subclass is not straight forward.
>>
>> I will provide such a subclass of AgentXSharedMOTableSupport for multi variable indexes and master/child tables within the next few days.
>
> Thank you, Frank, for this answer. Where will I be able to find this implementation? Because in my case, I have no idea what specific handling would be required, as suggested for this subclass.
>
> Kind regards,
> ch
>
>
>
> Am 02.03.2012 08:26, schrieb HAAS Christian:
>> Hello there!
>>
>> I have one of those "Hardly an understanding of the underlying system, but have to make it work" cases and I hope to find the right pointers here.
>>
>> Setup and Goal:
>> We are using SNMP4J together with Net-SNMP for our processes. On a host with Net-SNMP, we have several of our processes running and each of them shall report some internal statistics (e.g. amount of threads) into a shared table.
>>
>> This table has an index consisting of two columns: The unique process (component) ID and the statistics (entry) ID.
>> The idea is to access such a statistics row by using an OID
>> table.componentId.entryId
>>
>> Several of the entry IDs are reported by more than one process, some are reported by only a selected few - still, each in his own row though. All rows of ones own componentId are handled by the respective process.
>> For example:
>> table.componentA.numberOfThreads = ...
>> table.componentA.packetsReceived = ...
>> table.componentB.numberOfThreads = ...
>> table.componentB.packetsReceved = ...
>> table.componentB.numberOfSessions = ...
>>
>>
>> The Problem(s):
>> Filling in the expected values for the index data, each process can
>> add at most one row. (Typically, only one row would show up as for
>> second problem) Filling a unique 'counter' value into the componentId lets the processes add specific entries only once (e.g. only one is to report numberOfThreads; The first process to register reports all his entries, any further reports only those entryIds not yet used) Filling unique counters for both index fields, all processes can add all data.
>>
>> But the second two are probably not what is expected when reading the MIB.
>> The intent is to use the indexDef as a combined primary key, but as it appears, the two are handled being primary keys each on their own.
>>
>> MIB:
>> -- 1.3.6.1.4.1.3043.6.7.2.52
>> frqChNvpTable OBJECT-TYPE
>> SYNTAX SEQUENCE OF FrqChNvpEntry
>> MAX-ACCESS not-accessible
>> STATUS current
>> DESCRIPTION
>> "Name-value-pairs"
>> ::= { frqComponentHealthObjects 52 }
>>
>>
>> -- 1.3.6.1.4.1.3043.6.7.2.52.1
>> frqChNvpEntry OBJECT-TYPE
>> SYNTAX FrqChNvpEntry
>> MAX-ACCESS not-accessible
>> STATUS current
>> DESCRIPTION
>> "Name-value-pair"
>> INDEX { chNvpComponentId, chNvpEntryId }
>> ::= { frqChNvpTable 1 }
>>
>>
>> FrqChNvpEntry ::=
>> SEQUENCE {
>> chNvpComponentId
>> INTEGER,
>> chNvpComponentName
>> DisplayString,
>> chNvpEntryId
>> INTEGER,
>> chNvpEntryName
>> DisplayString,
>> chNvpEntryLongName
>> DisplayString,
>> chNvpEntryValueType
>> INTEGER,
>> chNvpEntryValueInt
>> Integer32,
>> chNvpEntryValueString
>> OCTET STRING
>> }
>>
>>
>> Code:
>> In code, a DefaultAgentXSharedMOTable is used, with the indexDef set up hopefully right:
>>
>> { // setup combined index
>> /** Sub indexes (SNMP)*/
>> final MOTableSubIndex[] subIndexes = new MOTableSubIndex[2];
>>
>> subIndexes[0] =
>> new MOTableSubIndex( new OID( MOTableBuilder.OID_CH_NVP_TABLE )
>> .append( MOTableBuilder.MO_COLUMN_IDX_CH_NVP_COMPONENT_ID ),
>> SMIConstants.SYNTAX_INTEGER, 1, 1 );
>> subIndexes[1] =
>> new MOTableSubIndex( new OID( MOTableBuilder.OID_CH_NVP_TABLE )
>> .append( MOTableBuilder.MO_COLUMN_IDX_CH_NVP_ENTRY_ID ),
>> SMIConstants.SYNTAX_INTEGER, 1, 1 );
>> this.indexDef = new MOTableIndex( subIndexes );
>> }
>>
>>
>> Questions:
>> 1. Is the basic goal possible at all? If not, I can save some problem finding and we need to go back to the drawing board.
>> 2. If possible, what am I doing wrong in the code? How should I do it?
>> 3. Is there anything more I should provide so I can help you help me
>> help us all? :) 4. Side question: What are the two last parameters of the MOTableSubIndex constructor for? It works only using 1, 1.
>>
>> I realize just days ago there seems like to have been a similar problem with a combined index on this list, but I can't see whether this is the same problem nor did I understand the proposed solution.
>>
>> Thank you and kind regards,
>> ch
>> ____________________________________________________
>> Christian Haas
>> Software Engineer
>> FREQUENTIS AG
>>
>> Innovationsstraße 1, 1100 Vienna, Austria
>> Phone +43-1-811 50 - 8353
>> Mobile +43-664-60 850 - 8353
>> Fax +43-1-811 50 - 77 8353
>> Web www.frequentis.com
>> E-Mail christian.haas at frequentis.com<mailto:christian.haas at frequentis.com>
>>
>> Handelsgericht Wien (Vienna Commercial Court): FN 72115 b DVR 0364797,
>> ATU 14715600 ____________________________________________________
>> Diese E-Mail könnte vertrauliche und/oder rechtlich geschützte Informationen enthalten. Wenn Sie nicht der richtige Adressat sind oder diese E-Mail irrtümlich erhalten haben, informieren Sie bitte sofort den Absender und vernichten Sie diese E-Mail. Das unerlaubte Kopieren sowie die unbefugte Weitergabe dieser E-Mail sind nicht gestattet.
>> This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorised copying, disclosure or distribution of the material in this e-mail is strictly forbidden.
>>
>>
>>
>> _______________________________________________
>> 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
>
> _______________________________________________
> 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++
Maximilian-Kolbe-Str. 10
73257 Koengen, Germany
https://agentpp.com
Phone: +49 7024 8688230
Fax: +49 7024 8688231
_______________________________________________
SNMP4J mailing list
SNMP4J at agentpp.org
http://lists.agentpp.org/mailman/listinfo/snmp4j
More information about the SNMP4J
mailing list