Zenperfsnmp error: zenperf is adding an extra .0 on my OID causing it not to read values

[everythingzen]

everythingzen [http://community.zenoss.org/people/everythingzen] created the discussion

"Zenperfsnmp error: zenperf is adding an extra .0 on my OID causing it not to read values"

To view the discussion, visit: http://community.zenoss.org/message/73936#73936

--------------------------------------------------------------
So I have some RH devices which are not reading essentially CPU/load etc.  So the things devined in the /Device/Server/Linux template.

In my zenperfsnmp log I see Error zen.zenperfsnmp: SNMP get returned empty value: servername.domain 1.3.6.1.4.1.2021.11.9.0.0  Should be ssCpuUser.0

From the Monitoring Template if I edit the data source it shows its value as 1.3.6.1.4.1.2021.11.9.0 and if I run Test against a Device and put in my devices that are failing..I get the right value and I don't see the extra 0 on the end.

If I run snmpwalk using 1.3.6.1.4.1.2021.11.9.0  the correct value comes out.  So why, in zenperfsnmp is the extra 0 getting added and how do I stop it?


I am running zenoss 4.2.3, NetSnmp 5.5.0

Now the Redhat devices it is collecting on are Redhat ES 4 Nahant Update 6.  and I do have some of the devices in the same device class that do work..so I'm clueless why some appear to be putting this extra 0 on it.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/73936#73936]

Start a new discussion in zenoss-users at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]

[Vaibhav.Narula]

Vaibhav.Narula [http://community.zenoss.org/people/Vaibhav.Narula] created the discussion

"Re: Zenperfsnmp error: zenperf is adding an extra .0 on my OID causing it not to read values"

To view the discussion, visit: http://community.zenoss.org/message/73959#73959

--------------------------------------------------------------
there is bug in the version of netsnmp

https://bugzilla.redhat.com/show_bug.cgi?id=261601 https://bugzilla.redhat.com/show_bug.cgi?id=261601
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/73959#73959]

Start a new discussion in zenoss-users at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]

[everythingzen]

everythingzen [http://community.zenoss.org/people/everythingzen] created the discussion

"Re: Zenperfsnmp error: zenperf is adding an extra .0 on my OID causing it not to read values"

To view the discussion, visit: http://community.zenoss.org/message/73963#73963

--------------------------------------------------------------
Thanks for the response.  That bug however does not address my problem.  It talks about values not being updated..my problem is zenperfsnmp

When Zenperffsnmp runs it is sending a request using 1.3.6.1.4.1.2021.11.9.0.0   when it should be using 1.3.6.1.4.1.2021.11.9.0 (Notice missing zero).  It does this on all of my RH4 devices.   It definetly does it with 5.1.2 installed on the client machines.  This applies to all OIDs requested from the /Devices/Server/Linux Template.. so the CPU, Memory, Load stuff.  Strangly my Interface (so ethernet stuff) are working fine for all RH4 boxes.

I upgraded some of them to 5.3 which is about as new of a version as I can get and it seemed to fix some of the boxes but not all. 

I was hoping to not have to but I guess I should start looking at the template code to see if there is something in there.  My only thoughts are the MIB tree is different for the RH4??  not sure.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/73963#73963]

Start a new discussion in zenoss-users at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]

[jmp242]

jmp242 [http://community.zenoss.org/people/jmp242] created the discussion

"Re: Zenperfsnmp error: zenperf is adding an extra .0 on my OID causing it not to read values"

To view the discussion, visit: http://community.zenoss.org/message/73976#73976

--------------------------------------------------------------
This sounds like one of the built in Zenoss code paths mis-figuring if your OS requires the insertion of a .0 or not to work. Searching the forums about that should find many threads - though with 4.x I'm not really sure what the work-around or fix would be.

--
James Pulver
ZCA Member
CLASSE Computer Group
Cornell University
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/73976#73976]

Start a new discussion in zenoss-users at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]

[everythingzen]

everythingzen [http://community.zenoss.org/people/everythingzen] created the discussion

"Re: Zenperfsnmp error: zenperf is adding an extra .0 on my OID causing it not to read values"

To view the discussion, visit: http://community.zenoss.org/message/73975#73975

--------------------------------------------------------------
I did try to search the forums first.  I either didn't use the right search terms or I didn't really see anything that was similar.  If you have a suggestion of what to search for that would be great.

Yea that is what I was thinking the problem is.  I suppose as a hack I could create a template that lops off the .0 so when zenperfnsmp adds the extra it only uses .0 not .0.0  Although I know there are some OIDs that need to end in a non 0.. so that wouldn't help those.  ie...1.2.34.0 where the template defines the OID as ...1.2.34  I'm not sure either how I get around the slap another 0 on the OID

What I don't get is why it puts the extra 0 on in the first place.  I know the difference that snmpget needs at least the 1 zero..but if your template is configured properly the data points should already have the .0 on it.

I suppose my other option is a cmd collector rather than snmp..
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/73975#73975]

Start a new discussion in zenoss-users at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]

[Shane Scott]

Shane Scott [http://community.zenoss.org/people/hackman238] created the discussion

"Re: Zenperfsnmp error: zenperf is adding an extra .0 on my OID causing it not to read values"

To view the discussion, visit: http://community.zenoss.org/message/73994#73994

--------------------------------------------------------------
Everything:

Honestly I'd update net-snmp on those boxes. An overriden template will work as well, but it's a messy fix.

Best,
--Shane Scott (Hackman238)
http://shanewilliamscott.com http://shanewilliamscott.com
http://linkedin.com/in/shanewilliamscott http://linkedin.com/in/shanewilliamscott
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/73994#73994]

Start a new discussion in zenoss-users at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]

[everythingzen]

everythingzen [http://community.zenoss.org/people/everythingzen] created the discussion

"Re: Zenperfsnmp error: zenperf is adding an extra .0 on my OID causing it not to read values"

To view the discussion, visit: http://community.zenoss.org/message/74004#74004

--------------------------------------------------------------
I've updated them to net-snmp-5.3.0.1-1.el4 but it seems some of the machines still have issues. Which was the latest version (rpm) for rhel 4 I could find.  I'll have to see if I can build 5.7 from source on my el4 boxes.  I'll give that a go first and if not I'll have to move toward custom templates since I don't know the magic flag zenperfsnmp is deciding to add the extra 0.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/74004#74004]

Start a new discussion in zenoss-users at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]

[Shane Scott]

Shane Scott [http://community.zenoss.org/people/hackman238] created the discussion

"Re: Zenperfsnmp error: zenperf is adding an extra .0 on my OID causing it not to read values"

To view the discussion, visit: http://community.zenoss.org/message/74007#74007

--------------------------------------------------------------
Everything,

Can you compare the snmpd config on a working el4 and non-working el4 box? I'm wondering if something is ary since they should act the same.

Best,
--Shane Scott (Hackman238)
http://shanewilliamscott.com http://shanewilliamscott.com
http://linkedin.com/in/shanewilliamscott http://linkedin.com/in/shanewilliamscott
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/74007#74007]

Start a new discussion in zenoss-users at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]

[everythingzen]

everythingzen [http://community.zenoss.org/people/everythingzen] created the discussion

"Re: Zenperfsnmp error: zenperf is adding an extra .0 on my OID causing it not to read values"

To view the discussion, visit: http://community.zenoss.org/message/74019#74019

--------------------------------------------------------------
I'm 80% sure they are the same.  I scpd the conf file from one to another.  I'll do a diff for snmpwalks though and see if there is anything else and/or if there are any other snmp rpms installed that might affect things.

Thanks for the response.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/74019#74019]

Start a new discussion in zenoss-users at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]

[everythingzen]

everythingzen [http://community.zenoss.org/people/everythingzen] created the discussion

"Re: Zenperfsnmp error: zenperf is adding an extra .0 on my OID causing it not to read values"

To view the discussion, visit: http://community.zenoss.org/message/74066#74066

--------------------------------------------------------------
Ok..so all RHEL 4 boxes have the same version of net snmp now.  the Conf files are the same..however one thing that is different is the non working boxes are using snmp proxy to get their snmp data.  They are behind a firewall so one of the accessable RHEL boxes acts as a snmp proxy.  Which I agree does complicate things.  My original problem I think is actually caused by multiple things. 

One.. net-snmp 5.1 just doesn't work well with at least this version of zenoss.  So upgrading to 5.3 fixed those..

Now previously as I was going through some of my old notes I wrote using 5.4 should be the min version I would of used.  I've built zenoss collection through snmp proxy before but those machines were newer and I was using zenoss 3.X back then.

However..even with that it is really zenperfsnmp that is sending the wrong OID..So it is almost like that zenpack for /Devices got smarter but dumber when it comes to assigning it the extra 0..hmm..

Well I am going to try and build a 5.4 version of net-snmp if that doesn't work I suppose I'll just make the template for the half a dozen machines this affects.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/74066#74066]

Start a new discussion in zenoss-users at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]

[everythingzen]

everythingzen [http://community.zenoss.org/people/everythingzen] created the discussion

"Re: Zenperfsnmp error: zenperf is adding an extra .0 on my OID causing it not to read values"

To view the discussion, visit: http://community.zenoss.org/message/74077#74077

--------------------------------------------------------------
Ok..I fixed it.  So just to follow up in case anyone else has this problem.

So I built net-snmp ver 5.4 for RHEL4.  I installed on all devices..this did not fix my problem.

I turned zenperfsnmp in DEBUG mode. I noticed on the affected boxes during the "Fetching OID chunk size....."  were not collecting the OIDs defined for CPU/Mem..the one from the /Devices template.  so the 2021...ones.

To recap this appears to be a problem on the devices that were using a SNMP Proxy.  Older versions of Zenoss did not have any issues with this.

Hmm.. So I went to the zope interface for the indifidual devices and there is a snmpindex field under the Device properties.   By default it is set to 0.  I deleted it and poof the Fetching OIDs now list the missing Device ones and I now have data.  Changing this also had no affect on the other graphs that worked previously (the ethernet ones for example).

So I don't know why exactly this was happening.  My only guess is somehow Zenoss already knows that I've collected the cpu/mem etc OIDs and drops them...for the proxy to work (collect devices behind a firewall using one machine as the snmp proxy) I have to use that one machines IP address for the collection of all of the proxied machines.  Which..you have to change my hand through zope. 

So it is working..I am not 100% happy with the solution but I'll have to debate if it is worth my time to trudge through the python code to see where it decides to either change the OID and slap on the extra 0 or why i doesn't fetch the OID from the templates.
--------------------------------------------------------------

Reply to this message by replying to this email -or- go to the discussion on Zenoss Community
[http://community.zenoss.org/message/74077#74077]

Start a new discussion in zenoss-users at Zenoss Community
[http://community.zenoss.org/choose-container!input.jspa?contentType=1&containerType=14&container=2003]