Result from SNMP not valid. Partial Result: U...

[krispal]

All what is by default works fine, that is host mibs. Tried informant-s oid-s as well - worked as well.

Tried to graph MSSQL User Connections oid ( .1.3.6.1.4.1.311.1.4.1.1.2.1.16) nothing is graphed and the error message in log CACTID: Poller[0] Host[1] WARNING: Result from SNMP not valid. Partial Result: U...

I use Getif to test OID-s - oid worked.

What can be the problem?

[BSOD2600]

Turn the polling level in cacti to debug (Settings | Polling level) for one cycle and paste the output of the log file (./cacti/cacti.log). If it contains WEBLOG entries, then don't bother posting since they just clutter up the log. Turn off WEBEVENTS logging and try again.

Look for the warning and see why it happened.

[krispal]

nothing helpful really...only these:

CACTID: Poller[0] Host[1] SNMP: v1: 127.0.0.1, dsname: snmp_oid, oid: .1.3.6.1.4.1.311.1.4.1.1.2.1.16, value: U

POLLER: Poller[0] CACTI2RRD: c:\rrdtool\rrdtool.exe update c:\Inetpub\wwwroot\cacti\rra\myserver_snmp_oid_161.rrd --template snmp_oid 1125035340:U

Changed the data source to some host oid - no errors, changed the oid with Getif (browsed the exact oid) to .1.3.6.1.4.1.311.1.4.1.1.2.1.17.3.65.68.83 (sql locks)

and no graphs

08/26/2005 10:20:14 AM - CACTID: Poller[0] Host[1] SNMP: v1: 127.0.0.1, dsname: snmp_oid, oid: .1.3.6.1.4.1.311.1.4.1.1.2.1.17.3.65.68.83, value: U

I found another thread about this http://forums.cacti.net/post-38700.html
No solution there either...

[krispal]

Actually, little bit searching revealed the cause -

-----------
This is a well known problem with the MS SQL server agent.

Microsoft disregarded it and recently we filed a service request with MS.

MS looked into it, verified it was a bug and claimed that they would NOT fix this behavior and WMI/WBEM was the way to SQL server would support monitoring in the future.
They said they would file a KB for it, I don't know if they have.
I have a feeling that Microsoft is slowly but steadily moving away from SNMP indicating SNMP's feeling security support as a prime reason.

GETNEXT by definition is a "lexicographic NEXT" object, so I do not think it is the correct way to perform "polling" for a specific object instance.

MRTG allows defining custom scripts to allow using some other data acquisition mechanism (other than SNMP).

Maybe you should write some perl script/vbscript etc that could poll the WMI object rather than the SNMP object. Maybe someone out there could share such a script to users like us....
-------------

So this problem is not related with cacti in any way.

But thank You for your time anyway:)

[BSOD2600]

WMI scripts already exist for cacti. Only downside is Win32 hosts can only run them. On the cacti additional scripts page, check out the w32_query_OperatingSystem.pl script. It has WMI calls, which you should be able to easily modify to suit your needs.