snmp dynamic indexes

[xs-]

Does the dynamic index support indexes of more than 1 number?
i.e.
NetApp Qtree /quota stuff uses 2 numbers for its indexes
Qtree name, base OID: 1.3.6.1.4.1.789.1.4.5.1.10
Qtree name, example index: 1.3.6.1.4.1.789.1.4.5.1.10.6.5

quota id, base oid:1.3.6.1.4.1.789.1.4.5.1.4
quota id, example index: 1.3.6.1.4.1.789.1.4.5.1.4.6.5

Will this work? I keep getting 'no such oid' =\

[Alexei]

I am afraid this will not work, however I am not 100% confident.

[xs-]

Theoretically it could be made to work, the dynamic part of the oid is just the difference between the 'base search partial oid' and the found oid using the keyword (instead of just the last number in the oid).
Could this be put on the todo list?

[ggiesen]

I have the exact same problem as well. I have a 3-level index, and Zabbix finds the indexed item correctly, but does not correctly append the full difference between the indexed item you're searching for, and the item you want to find the value for.

Example (from CISCO VPDN MIB):

.1.3.6.1.4.1.9.10.24.1.3.2.1.2.2.24196.60898 = STRING: "[email protected]"
.1.3.6.1.4.1.9.10.24.1.3.2.1.3.2.24196.60898 = INTEGER: 10
.1.3.6.1.4.1.9.10.24.1.3.2.1.4.2.24196.60898 = Timeticks: (52950267) 6 days, 3:05:02.67
.1.3.6.1.4.1.9.10.24.1.3.2.1.5.2.24196.60898 = Counter32: 3685231
.1.3.6.1.4.1.9.10.24.1.3.2.1.6.2.24196.60898 = Counter32: 349105822
.1.3.6.1.4.1.9.10.24.1.3.2.1.7.2.24196.60898 = Counter32: 6645606
.1.3.6.1.4.1.9.10.24.1.3.2.1.8.2.24196.60898 = Counter32: 3877033828
.1.3.6.1.4.1.9.10.24.1.3.2.1.9.2.24196.60898 = INTEGER: 1
.1.3.6.1.4.1.9.10.24.1.3.2.1.10.2.24196.60898 = ""
.1.3.6.1.4.1.9.10.24.1.3.2.1.11.2.24196.60898 = INTEGER: 0
.1.3.6.1.4.1.9.10.24.1.3.2.1.12.2.24196.60898 = INTEGER: 2
.1.3.6.1.4.1.9.10.24.1.3.2.1.20.2.24196.60898 = Gauge32: 0
.1.3.6.1.4.1.9.10.24.1.3.2.1.21.2.24196.60898 = Counter32: 0
.1.3.6.1.4.1.9.10.24.1.3.2.1.22.2.24196.60898 = Counter32: 0
.1.3.6.1.4.1.9.10.24.1.3.2.1.23.2.24196.60898 = ""
.1.3.6.1.4.1.9.10.24.1.3.2.1.24.2.24196.60898 = INTEGER: 0

The item I'm using as my search is "[email protected]", but the base of the index is (according to the MIB) ".1.3.6.1.4.1.9.10.24.1.3.2.1.2". Therefore "2.24196.60898" is the index number. I'm trying to retrieve the value for ".1.3.6.1.4.1.9.10.24.1.3.2.1.6.2.24196.60898" , of which the base should be ".1.3.6.1.4.1.9.10.24.1.3.2.1.6".

Zabbix is reporting the following: "OID [1.3.6.1.4.1.9.10.24.1.3.2.1.6.60898] value has unknow type [0x81]". This should be pretty trivial to compare the difference between the the matched index ".1.3.6.1.4.1.9.10.24.1.3.2.1.2.2.24196.60898" , and the search base index we supply, and then append it to the index we're looking for the value for ".1.3.6.1.4.1.9.10.24.1.3.2.1.6"

I'd love to see this implemented as we have a lot of PPPoE sessions that we'd like to track that have constant usernames that obviously have changing indexes. This should be a relatively simple fix to implement, any chance it can be done?

Thanks

[ggiesen]

Sorry for the thread bump, but this is becoming an important issue for me. I'm about to take my zabbix install to production phase and this is a big unchecked box in terms of being ready. Can we expect to see a 1.6 release with this, especially since 1.8 has been pushed back to October?

[xs-]

Well, it should be a simple thing. We are basically talking about the difference between a base-oid and the 'actual' list gotten from that base-oid. Shouldn't really matter if the difference is 1 int or multiple seperated by dots (ok its not an int anymore then, but still).

just my 5 cents.

[jponeill125]

I know this was a long time ago. But I am in the same boat as ggiesen and was wonder wondering was a solution ever found. I am currently running 1.8.4.

[kingpin66]

another similar problem...

hi xs

Not an answer, but may you can help me with something similar...

I like to monitor the processor ussage on a virtual machine on a esxi hypervisor.
I work with SNMP, Windows server 2008 with up to 4 virtual cores.
Because of esxi loadbalancing, the first core is not always on index 1 of the OID "hrProcessorLoad".
But all cores are in a row, ae: indexes1/2/3/4 or indexes 3/4/5/6"
Also with dynamic indexes, i'm always just able to read the load of the first core. (all are same named in the devicedesc.)

My request is:
Can i somehow add a index offset to the dynamic index option? Ae: ["index" + 1,...,....]
I tryed amost all combinations with +1, but nothing seams to work.

Btw: I like to monitore only by snmp... and don't like to install the agend.

Cheers and thanks