Weblogic and SNMP Problems

[thorpa]

Hi,

I have been trying to get cacti to work with SNMP and Weblogic but am not having much luck and would like some suggestions from the experts as to what might be causing the problem. I am using Cacti 0.8.6f with snmp support and the standard "Data Input Method - Get SNMP Data". Perhaps this is the problem?

If I use snmpget to retrieve the OID information, it works fine, so I am simply using the same OID and inserting it into Cacti and running php.cmd to view the output as shown below.

Using snmpget
snmpget -c public -v 1 xxx.xxx.xxx.xxx enterprises.140.625.340.1.25.32.48.48.98.101.50.52.100.97.97.52.57.99.98.50.51.55.53.57.50.56.50.54.98.99.56.100.55.101.48.51.98.53

Result is good
SNMPv2-SMI::enterprises.140.625.340.1.25.32.48.48.98.101.50.52.100.97.97.52.57.99.98.50.51.55.53.57.50.56.50.54.98.99.56.100.55.101.48.51.98.53 = INTEGER: 35366312

When Using Cacti to get the result
08/10/2005 10:40:07 AM - CMDPHP: Poller[0] Host[4] DS[17] WARNING: Result from SNMP not valid. Partial Result:
08/10/2005 10:40:07 AM - CMDPHP: Poller[0] Host[4] DS[17] SNMP: v1: 134.251.199.133, dsname: snmp_oid, oid: enterprises.140.625.340.1.25.32.48.48.98.101.50.52.100.97.97.52.57.99.98.50.51.55.53.57.50.56.50.54.98.99.56.100.55.101.48.51.98.53, output: U

[pierre-luc]

Hi thorpa,
I just spend times trying to get some stats from weblogic using SNMP with cacti. I never be able to get valid data user snmp_queries directly from cacti, So. I think for now it's by using custom scripts.

The problem I get between weblogic+snmp and cacti is that indexes from snmpwalk are too longer for cacti.

[TheWitness]

Please see if you are having the following problem:

http://bugs.cacti.net/view.php?id=530

TheWitness

Please post both here and in the bug report either way. Thks!

[pierre-luc]

Hi TheWitness,
Yes, I got the bug you specify on the URL. the OID index is longer than 132 characters in weblogic.

Thanks.

[pierre-luc]

Because weblogic snmp community name is related with de managed-server name, I build a Perl script that go query all managed-server for JMS queue, It still working for our system.

There is the script that can run manually for testing and the Data Query XML for cacti.


The script is a bit scrappy but it work, I'not a good perl scripter.

[TheWitness]

I believe you can alter the table's per the bug report. We will move this to the 0.9 version as well.

TheWitness

[thorpa]

Hi,

The characted count for the OID is 132 in total, so this does not appear to be the problem.

The SNMP Cache shows this.

Action: 0, OID: enterprises.140.625.340.1.25.32.48.48.98.101.50.52.100.97.97.52.57.99.98.50.51.55.53.57.50.56.50.54.98.99.56.100.55.101.48.51.98.53 (Host: 134.251.199.133, Community: public)

Which reflects the input OID
enterprises.140.625.340.1.25.32.48.48.98.101.50.52.100.97.97.52.57.99.98.50.51.55.53.57.50.56.50.54.98.99.56.100.55.101.48.51.98.53

I might make the changes to the Database anyway and see how this goes, else I might try the script from pierre-luc

I'll let you know how it goes.

[TheWitness]

Perform the database changes. The previous width was 100 characters and not 132. I am including this in the next maintenance release of Cacti.

TheWitness

[thorpa]

I made the changes to the DB as follows

mysql> ALTER TABLE `host_snmp_cache` CHANGE `snmp_index` `snmp_index` VARCHAR(255) NOT NULL;
Query OK, 18 rows affected (0.04 sec)
Records: 18 Duplicates: 0 Warnings: 0

mysql> ALTER TABLE `data_local` CHANGE `snmp_index` `snmp_index` VARCHAR(255) NOT NULL;
Query OK, 7 rows affected (0.00 sec)
Records: 7 Duplicates: 0 Warnings: 0

mysql> ALTER TABLE `graph_local` CHANGE `snmp_index` `snmp_index` VARCHAR(255) NOT NULL;
Query OK, 7 rows affected (0.00 sec)
Records: 7 Duplicates: 0 Warnings: 0

but I get the same problem, so I might try an alternative method.

08/12/2005 06:31:57 AM - CMDPHP: Poller[0] Host[4] DS[18] WARNING: Result from SNMP not valid. Partial Result:
08/12/2005 06:31:57 AM - CMDPHP: Poller[0] Host[4] DS[18] SNMP: v1: 134.251.199.133, dsname: snmp_oid, oid: enterprises.140.625.340.1.25.32.48.48.98.101.50.52.100.97.97.52.57.99.98.50.51.55.53.57.50.56.50.54.98.99.56.100.55.101.48.51.98.53, output: U

[TheWitness]

You should delete and then re-add your host. If you have a lot of data, then you should delete the affected graphs and data sources and then recreate.

TheWitness

[TheWitness]

Also, please remove the word "enterprises" from your query. Use the "pure" OID.

TheWitness

[mvam]

thorpa

also note that weblogic has dynamic OIDs. so each time you restart your server your OIDs will change. i wrote a perl script to deal with the changes. i can post it if you like.