SNMP Host Selection Help - Page 3


You specify seven things on this page:

If you wish, enter an alias name ("human-readable") for this host. This will be used when displaying reports, but not when selecting hosts on the index page. If left blank, it will attempt to use the "system.sysName" SNMP entry for it. If that fails, whatever you put in as its hostname (or IP address) will be used.

Turn on (select) the "Dynamic Collection" checkbox to enable getting around dynamic interface numbering on this host. Some SNMP agents renumber their interfaces at the most inopportune times, causing lesser statistics collectors to associate stats with the wrong interfaces. This "dynamic" option will force NISCA to, every time it fetches data from this host, re-read the interface name-to-ifIndex mapping so this won't happen. Only enable this if you're positive this host rearranges its interface numbers now and then, because it substantially slows down data collection from that host. You might even need to adjust the global "Delay" setting if this has a negative-enough impact.

Turn on (select) the "On" checkbox next to each interface you want to monitor.

If a host/interface supports SNMPv2 (64-bit counters) and you want to use them on it, leave on (selected) the "64" checkbox next to it. Note that you cannot do this on interfaces that do not return a value for the SNMP entry "ifHCInOctets"; there will be no "64" checkbox on such interfaces. Some hosts have ifHC__ entries but refuse to return data on them; Cisco 2900-series switches are one example. You can't use SNMPv2 on them for this reason; there'd be no point. Interfaces which do support it will appear on this page in this color, as well as having a checkbox which comes pre-enabled; otherwise, a big "X".

Note: Just enabling 64-bit counters on an interface will not fix any existing statistics that were collected from it using 32-bit counters. Once it's been collected, there's no helping it. :(

For each interface, you can specify an alias name to make your reports more understandable.

The "CIR" entry for an interface will be used to draw a nice line across your graph to show you whether your traffic is bursting above it or not. Graphs will not be expanded vertically just so the CIR level will be included on them; if you put in a CIR of 9 mBps and you never get above 8 mBps, you'll never see the CIR line across a graph for that interface. Enter it in bytes per second (e.g., a CIR of 64 megabits per second should be entered as "8000000" or "8,000,000"). Any non-numbers in the CIR entry will be stripped out, meaning you can put commas in it if that comforts you. :)

The "ifSpeed" entry is done as follows: If there's a speed entry in the database for an interface, it will put that into the textbox. If not, it will attempt to retrieve the current ifSpeed entry on it via SNMP. If that succeeds, that value will be put in the text box; if not, it will be set to zero. If the ifSpeed reported by SNMP differs from that in the database, the SNMP-reported value will be displayed under the textbox so you know it happened. You really should make sure none of the interfaces you're monitoring has a zero ifSpeed entry; it can cause ugly things to happen on the graphs if you switch from SNMPv1 to v2c (or vice versa) on it. You can use commas in this field, too, to make the numbers more sane to look at.

Please note... Just because an interface does SNMPv2 doesn't mean there's a 64-bit counter enabled for it; that's why you'll probably see some ifAlias entries show up on interfaces for which you can't enable SNMPv2.

As usual, when you have all the options set the way you want, click the "Go" button.


Main Help Index