SNMP Traps with UPS

[evensteven]

We've used HostMonitor for SNMP before but always when an OID was involved (routers, switches, ...).
Now we have a BestLink SNMP/Web Adapter (http://www.powerware.com/UPS/Connectivity/Legacy.asp) set up for monitoring our UPS. These come with MIB files.
I started by creating a test SNMP Trap but am at a loss as to how the MIB files fit within the Bad Message Filter and Good Message Filter fields. The only one that appears obvious is of course the SNMP Device (address).
I've searched the forums and the web for examples, I've read through this sites documentation, but still can't make my way through this.

Unfortunately Powerware makes one register first before viewing their MIB files, but I can post whatever is required. The entire file is 2255 lines long and therefore way to long to post completely.

Thanks for reading.

[timn]

I'm not familiar with the Powerware UPS but we do use Host Monitor (HM) to monitor 3 APC Symmetra UPS units at this facility.

First, the MIB your UPS vendor provides is simply a 'definition' of the data that your Powerware unit is capable of providing 'if asked' via an SNMP request from a remote client. A MIB defines the valid OIDs (e.g. 3.1.3.6.156.7.1.2) that may be requested from your Powerware UPS.

For example, at our facility, I can look at the Output Current (in amps) by having HM request the OID 1.3.6.1.4.1.318.1.1.1.4.2.4.0 from our Symmetra.

Next, there are two separate SNMP 'test methods' for using HM to communicate with your UPS. The first is as I've explained above -- HM queries your UPS at regular intervals, asking for specific OIDs. This equates to the "SNMP" test method in HM.

SNMP traps work the other way around - when a 'significant event' occurs on the UPS, the UPS sends a packet to HM (without being 'asked' by HM). The "SNMP Trap" test method in HM simply establishes a "listener" that waits for incoming messages from your UPS.

We happen to use a single instance of the "SNMP Trap" method for all possible traps from all possible devices. When HM receives a trap from a device, it simply emails the OID and the Value.

Configuring what events and conditions trigger a trap -- i.e. the things that cause your UPS to send a message to HM -- is usually done from the web/html interface of your UPS. Configuring a trap will also necessarily involve telling your UPS about the IP address of your Host Monitor application.

You can often (but not always) ignore the MIB when configuring SNMP traps. It's often good enough to know you'll get a trap when an event occurs -- since there is often text explaining the event included with the SNMP packet. In such circumstances, knowing the exact OID is often not important.

Contrast this to the first method of SNMP communication where you must know the OID before you can ask for what you want.

From here, it's just a matter of knowing what you want to monitor and whether you want HM to poll for the data or you want your UPS to trap for the event.

We happen to use a combination of both methods at our facility. For example, we poll each of the UPS units for the following (using HM's "SNMP" test method):

Output Current
Output Load
Battery Run Time
Battery Temp
Input Line Voltage
Output Load
Output Status (On Battery, On Line)

There are over a hundred different events that are trapped for by our UPS -- most of which never occur. Here is a partial list to give you a feel (uses HM's "SNMP TRAP" test method):

0x011B UPS: In bypass due to an internal fault. Severe
0x011C UPS: In bypass due to user command via software or panel. Warning
0x011D UPS: In bypass initiated by user. Warning
0x011F UPS: Base module bypass power supply failure. Severe
0x0120 UPS: Base module fan failure. Severe
0x0122 UPS: External battery pack communications lost. Severe
0x0126 UPS: AVR Boost or AVR Trim relay failure. Warning
0x0128 UPS: Bad output voltage condition. Warning
0x012A UPS: Battery charger failure. Warning

[evensteven]

Thank you for your reply timn.

I had not realized when initially writing that the 1.3.6.1.4.1 was a universal ID. I noticed this comparing what we currently had with our Cisco and your posing for the UPS.
Adding this to their MIB file:
bestPower OBJECT IDENTIFIER ::= { enterprises 2947 }
bestLink OBJECT IDENTIFIER ::= { bestPower 1 }

upsIdent OBJECT IDENTIFIER ::= { bestLink 1 }
upsBattery OBJECT IDENTIFIER ::= { bestLink 2 }
<snip>
upsIdentModel OBJECT-TYPE
SYNTAX DisplayString (SIZE (0..29))
ACCESS read-only
STATUS mandatory
DESCRIPTION
"The UPS model name (example: 'FE500VA')."
::= { upsIdent 2 }
<snip>

So taking that information I ended up with:
1.3.6.1.4.1.2947.1.1.2.0
Correctly returning FE7.0KVA

I would have appreciated Powerware including the initial OID or an example within their file or documentation, but apparently I need to read up on SNMP.

Thanks again.

[timn]

Just FYI, I use a tool called OIDView to explore our devices via SNMP and then copy out specific OIDs for use in HM.

OIDView is a commercial tool and fairly powerful -- they've also got a huge online MIB database where you can get MIBs for just about anything.

If the cost is a bit high, there are also a number of freeware/shareware utilities available that perform a similar function.

BTW, here's an interesting article on the history of the various versions of SNMP.

[evensteven]

Thanks again for the replies and for the URLs.
The online MIB database is bookmarked in the more than probable possibility that we start to monitor more devices on our network.