Interface Index Problem

jromariz · Post by **jromariz** » Tue Aug 26, 2003 6:41 am

Hi Alex,

Don´t know if it´s a problem or I´m missing some info. We use HM mainly do monitor the status of hundreds of routers subinterfaces and get informed when any of them goes down. The problem I´m facing is that whenever I change the interfaces (add, remove, etc.) their index number (which is apeended to the OID number) also changes. This is a very big problem for us because we have to check all the tests on the same router and correct their SNMP instance. In MRTG it is possible to use the IP address or descriptio associated with each interface to perform the tests, like this there´s no problem when the interfaces´s index changes.

Hope you understand my problem.

Thanks,

Jromariz.

KS-Soft · Post by **KS-Soft** » Wed Aug 27, 2003 1:12 am

Yes, I understand what is the problem. But I am not sure I understand how to avoid it. Probably MRTG requests router's table of interfaces, gets interface index by IP address or description, and then makes another request.
Its possible but not flexible enough for generic SNMP test. Create new test methods to resolve this issue is not great idea as well. Probably some macro variables that will tell to perform another SNMP request and use retrieved value as value of the variable...

Regards
Alex

jromariz · Post by **jromariz** » Tue Aug 17, 2004 6:37 am

Hi Alex,

Just to ask if at this time theres any solution to this problem, I´m still having a lot of work everytime a reboot my routers.

Regards,

Jromariz.

hamoja · Post by **hamoja** » Tue Aug 17, 2004 6:50 am

If your sub interfaces are configured with an IP address, can you monitor by IP vs by SNMP?. Either way should report up/down status unless you are using SNMP for other special configurations.

jromariz · Post by **jromariz** » Tue Aug 17, 2004 6:53 am

HI Hamoja,

I believe it will work. I´ll give it a try.

Thanks for helping,

Jromariz.

meppyman · Post by **meppyman** » Sun May 02, 2010 6:31 am

Just bringing this one up again as it is an issue for me now that I am refining my Host Monitor setup and wanting to monitor more things. I want to use SNMP to check link status, error count, interface speed, etc. But because the index changes each time I am stuck.

It would seem that the only way to manage this would be to refine the way HM SNMP Get works to do a lookup to get the index (based on ifDescr perhaps).

Any thoughts on how to work around it without waiting for HM to support this feature?

meppyman · Post by **meppyman** » Sun May 02, 2010 6:42 am

Looks like maybe I should use the Traffic Monitor for most of this? Will play around with that to see if it can do everything I need.

EDIT: looks like it has the same problem when the index changes. So I am stuck again.

NB: I am wanting to monitor network interfaces on the server primarily, but would be nice to monitor other interfaces as well.

KS-Soft · Post by **KS-Soft** » Mon May 03, 2010 12:48 pm

Yeah, I do not see how its possible to setup SNMP test when you don't know what exactly counter you need to check.
Its like tell HostMonitor to check all counters between 1.3.6.1.2.1.2.2.1.2.1 and 1.3.6.1.2.1.2.2.1.2.9999, find counter that contains specific string (e.g. 'interface name 2'), get index of this counter then use this index to construct OID 1.3.6.1.2.1.2.2.1.7.<index> and finally check value of 1.3.6.1.2.1.2.2.1.7.<index> counter.

Why your device changes indexes of its interfaces all the time? May be there is some option to prevent this from happening?

Regards
Alex

meppyman · Post by **meppyman** » Mon May 03, 2010 10:29 pm

I always thought there was a way to do a table lookup for these with SNMP?

Not sure why they change, only just discovered the issue the other day trying to monitor them for the first time. They are standard NICs on HP servers under Windows 2003. They change if you disable and re-enable or power off and on again the interface or server.

It seems that the problem happens on other devices to like smart switches. Other monitoring applications seem to handle it, not sure how they do it. It's annoying for both you and I

.

EDIT: this is from Wikipedia...

Index assignments for an SNMP device instance may change from poll to poll mostly as a result of changes initiated by the system admin. If information is needed for a particular interface, it is imperative to determine the SNMP index before retrieving the data needed. Generally, a description table like ifDescr will map a user friendly name like Serial 0/1 (Blade 0, port 1) to a SNMP index.

meppyman · Post by **meppyman** » Mon May 03, 2010 10:39 pm

Also found this from Cisco, seems like there may be a way to set persistence, I am pretty new to SNMP so have no idea yet how to set values and what SNMP tools I should be using, etc.

KS-Soft · Post by **KS-Soft** » Tue May 04, 2010 5:20 pm

I always thought there was a way to do a table lookup for these with SNMP?

Yes, you can walk thru counters provided by SNMP agent find desciption then request another conter and so on.
But when monitoring application have to make 30 requests instead of single request, its not a efficient way. When you need to monitor 300 routers and check 9,000 interfaces, then its better to perform 9,000 requests, not 9000x30=270,000 requests.

BTW when you need to check total traffic on all enabled interfaces, its not a problem. In such case you (and HostMonitor) don't need to know index of some specific interface, it checks all available interfaces of the device. So in such case Traffic Monitor test works fine even if device changes order/indexes of its interfaces.

Regards
Alex

meppyman · Post by **meppyman** » Tue May 04, 2010 7:23 pm

Of course, I understand... but I am not sure I understand why you say 30 requests when I can get the new indexes with one or two requests from MIB Browser?

I am not sure why I feel like I am defending things here, it is obviously a design of SNMP and you are the expert not me

I just want to know how to work around it... do you understand it enough to suggest how I might be able to stop the indexes changing or what a work around might be? I will also ask on HP forums but unsure if other devices will also have the problem?

KS-Soft · Post by **KS-Soft** » Wed May 05, 2010 11:20 am

30 - is just example. Some tables may contain just 1 item, other may have 1000 items.

do you understand it enough to suggest how I might be able to stop the indexes changing

I don't think there is some common option that allows you to do this

It depends entirely on developers of SNMP agent software.

Regards
Alex

meppyman · Post by **meppyman** » Wed May 05, 2010 4:12 pm

So basically if the indexes change I am screwed then and cannot monitor those interfaces. Oh well at least I know where I stand, I will chase it up with the hardware vendors now.