I need monitoring 700 routers, but the bandwith is of only 256k for reach each router.
I am triying monitoring with snmpcollector and I see that the router it send timeout message, that is happen because snmpcollector probe discover all metric family.
For resolved this issue I think that i can setup snmpcollector for that only discover interface metrics. So I have some question
Is possible configure snmpcollector for that only discover interface?. Really us only interesting the metric operstatuspolleable
Thanks so much if anyone can help me
Yes, you can create a Rule with a Metric Family Filter in the Template Editor, as you can see more details at:
v2.0 snmpcollector AC GUI Reference - CA Unified Infrastructure Management Probes - CA Technologies Documentation
A point of clarification between discovery and monitoring.
Discovery is a two step process, first via discovery wizard to obtain basic device info, and second via snmpcollector which is based on the OIDs the device responds to. From those OIDs the profile resource tree is built.
Monitoring of the device is set via what resource items are enabled for QoS & Alarms. This can be set at the profile but better done via templates, and that is what will restrict the monitoring to only the desired items.
David / Alex. thanks for your comments.
The principal problem is that when discover a device, snmpcollector discover all metric family I need only discover Interface not all. Is possible this?
I am disable all metric familiy under snmpcollector in raw configure and only enable normalized-ports.xml
It is hard to re-configure the default scope of discovered components in snmpcollector probe.
If your business interest is only a few OID points, then you might consider SNMPGET probe, than snmpcollector probe.
In snmpcollector probe, I suggest that you use SNMP V1 for the device.
SNMP V1 is using single SNMP request, loads to target device becomes less heavier than V2/V3
If V1 is not allowed, use MAX_BATCH parameter to assign lower number of bulk requests.
Hope it helps.
Sorry but this just is not understood, why discovery is such a big issue. Are these devices being reconfigured constantly? If not why do they have to be rediscovered frequently?
Consider a large router for which the secondary discovery done by snmpcollector builds a profile with a thousand resource items. That effects the local database where the profiles are kept, and that's all.
You start out talking about the load by 'monitoring' and then focus on discovery. Discovery is an occasional event and monitoring is on going so it sounds like the need is to restrict 'monitoring' and not 'discovery'.
If a template is setup with only the one QoS you want to monitor then that's all that will be polled during each monitoring cycle, regardless of how many resource items are in the profile tree.
Thanks for your comments to all. We configured the metric only "normalizaed-port.xml" this kb helped https://comm.support.ca.com/kb/after-update-to-snmpcollector-v-32-many-device-profiles-in-failed-state/KB000045605 and this work for me.
We can not use SNMPv1 because this is a company banking and the security policy that we should use is SNMPv3.
We defined a template only with monitoring interfaces, and this working for me. Now is not estable because each cycle monitoring depending of the time of response of router. and maybe can get it the oid (operstatuspolleable) or maybe not, and it's a lottery in the router for the bandwidt short they have .
For solvent this problem, we define the monitoring with net_connect. We are Only having monitoring of status of interface up/down.