Thunderbolt failed to find APC network UPS

[Brian Bergin]

We have an APC HS-500 connected to our LAN that's web manageable that runs the power to our router/switches.  Thunderbolt failed to find this device even though it's pingable.  I was able to manually add it to the topography.

[Michael Holloway]

Hi Brian,

Can you tell us about the topology of the network?  VLans or multiple subnets?

-mike

[Marcos Hernandez]

Also Brian, with your permission, we could take a look at your site and help you determine why this is happening.

Thanks,

Marcos

[Brian Bergin]

Absolutely.  Let me know what I need to do.

[Marcos Hernandez]

Brian,

I created a monitoring rule (ICMP) to your APC UPS and it came up red immediately. Are you sure this device is on that IP you entered? It seems to be down...

Marcos

[Brian Bergin]

Absolutely.  See attached...  BTW, if this unit goes offline I have no Internet and no VoIP phones as it runs our RV082, DSL & cable modems, switches, the WAP4410N, ESW switch, etc...

[Brian Bergin]

I just found an "unknown" device with the same IP as the one I added for the APC HS-500 UPS.  The only difference I see is I didn't give it a MAC address but the portal found one.  Pings do work to that so the quesiton is why does the manually added one not work?  I'll leave both for now in case you want to see both.

[Michael Holloway]

Hey guys, this is an interesting corner case that I don't think we saw in testing for this initial release, so good find.  This would be where the device was discovered and labeled (as unknown in this case) and then a second host was added manually and given the same IP address as the first.  Go ahead and delete the manually added host, and feel free to rename and change the classification for the automatically discovered one.  We'll look to address this one in a future release.

-mike

[afullfor]

This UPS device is a little mysterious.  I've confirmed that the TBA doesn't get a ping response from it, as indicated by the monitor function.  The ping was, however, enough to trigger ARP resolution, which is what resulted in the unknown device appearing in the topology.  Discovery uses a UDP packet to solicit ARP as a way of finding devices on the LAN segment.  This was not triggering ARP responses for this device, although the ping did.  Once the device appeared in the ARP table it was added into discovery.  In fact, if you had entered a MAC address in your manually created entry, that would have been merged in with the ARP-derived information.

As for why the device doesn't answer ping from the TBA, but clearly does answer from your PC, it is hard to say.  Is it possible the device will only respond to the system running the management software?  Can you use a separate computer to see if it can reach the device?

As I say, this is rather mysterious.  The TBA at its heart runs Linux and the Linux ping command is not able to see a response.  If there is nothing filtering packets, I'm not sure where to look.  I have never encountered anything that answers ping to a Win PC and not to a Linux box (or vice versa).

[Michael Holloway]

Good catch Andy.  Merging the 2 hosts by supplying the MAC to the manual host would be a better recommended way to clean this one up as any monitors added get merged.  Sounds pretty mysterious, I'm expecting it turns out that the device just doesn't like to be solicited on that UDP port, or that an intermediate security device is blocking the UDP packet on ingress before causing an ARP who-has on the other side.

-mike

[Brian Bergin]

There isn't any management software for the HS-500, its management is purely web-based.  It's at the other end of the building and has no PCs connected to it.  It's designed purely as a wiring closet solution where there are no PCs and no people.  Basically, if a device hangs I can remote into the APC and power cycle one of the ports which of course forces a reboot.  The unit also provides UPS battery backup.  More on it can be found at:

http://www.apc.com/products/resource/include/techspec_index.cfm?base_sku=BH500NET&total_watts=200

BSBhttp://www.apc.com/products/family/index.cfm?id=3

[afullfor]

Brian, I'm still rather mystified about why our monitoring is not able to ping the APC but you can locally.  One thing I noticed is that Linux ping uses a payload of 56 bytes by default.  Windows ping uses 32 by default.  Could you try pinging the UPS with 56 bytes?  I think you use "ping -l 56". This is kinda clutching at straws here, but I don't have another explanation for what's going on.

[Brian Bergin]

F:\>ping 192.168.100.98 -l 56

Pinging 192.168.100.98 with 56 bytes of data:
Reply from 192.168.100.98: bytes=56 time=9ms TTL=100
Reply from 192.168.100.98: bytes=56 time=16ms TTL=100
Reply from 192.168.100.98: bytes=56 time=6ms TTL=100
Reply from 192.168.100.98: bytes=56 time=6ms TTL=100

Ping statistics for 192.168.100.98:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 6ms, Maximum = 16ms, Average = 9ms

[Michael Holloway]

Brian, did you do this ping from the same machine that was previously able to ping the APC, or a different host?

-mike

[Marcos Hernandez]

I found the following description in their documentation. Andy, maybe we need to force the ARP binding before we can get to it? Or maybe not, because it is showing on the TBA ARP table (and in discovery) it's just that we can't ping it. Could we try with 113 bytes? Read this:

From any computer on the same subnet as the Management Card, you can use ARP and Ping to assign an IP address to a Management Card, and then use Telnet to access its control console and configure the other TCP/IP settings.

After a Management Card has its IP address configured, you can use Telnet, without first using ARP and Ping, to access that Management Card.

1. Use the Management Card’s MAC address in the ARP command to define the IP address.

For the MAC address, look on the bottom of the Management Card or on the Quality Assurance slip included in the package.
For example, to define 156.205.14.141 as the IP address of a Management Card with 00 c0 b7 63 9f 67 as its MAC address., use one of the following commands:
– Windows command format:
arp -s 156.205.14.141 00-c0-b7-63-9f-67
– LINUX command format:
arp -s 156.205.14.141 00:c0:b7:63:9f:67

2. Use Ping with a size of 113 bytes to assign the IP address defined by the ARP command. For the IP address defined in step 1, use one of the following commands:
– Windows command format:
ping 156.205.14.141 -l 113
– LINUX command format:
ping 156.205.14.141 -s 113

3. Use Telnet to access the Management Card at its newly assigned IP address. For example:
telnet 156.205.14.141

4. Use apc for both user name and password.

5. See “Control console” on page 16 to finish the
configuration.