PowerChute has to use same netmask with AP9631

[Anonymous user]

This issue is extended out of "PowerChute cannot connect to management card" at http://forums.apc.com/spaces/5/smart-ups-symmetra-lx-rm/forums/general/1595/powershell-scripts-for-p...

For my configuration, the nodes in my cluster have to use different netmask. For example, the master node use 255.255.255.0 while the computational nodes use 255.255.255.128.

In this case, since the UPS (SMX3000RMLV2UNC) has to be shared with multiple nodes, it cannot be connected for both master node and a computational nodes.

I do not have a radius server.

Is it possible to configure the management card or the PowerChute to not require same netmask? Or is it possible for you to develop this option in your next update of firmware or PowerChute?

Thanks.

Version:
AP9631 firmweare: 5.1.5
PowerChute: 9.0.1

[Anonymous user]

I think I have a similar problem?

Our gateway is 10.8.16.1

But the nmc card is on a different private subnet, 10.8.20.9

The computers I want it to shutdown are on the 10.8.20. subnet. The DNS and email tests work with these settings:

IP 10.8.20.9
subnet 255.255.0.0
gateway 10.8.16.1

But PCNS registers it's IP with the NMC, but says it cannon communicate!

[BillP]

Hi Shane,

Can you ensure that ports 80, 3052 and 6547 on TCP and UPD are open.

[BillP]

What version of PCNS did you try to install? Correct me if I am wrong, the way I understand the problem is that during setup of PCNS it does a handshake with the NMC but after that once you log in you get a communication issue with PCNS and the NMC.

[Anonymous user]

the newest one, 2.2.4 I think?

Yes, it installs, and if I log into NMC I can see the IP of the machine I installed PCNS on. But when I open PCNS on the machine, it says it cannot communicate with the NMC

[BillP]

If you do have a Server type Operating Systems (Windows 2008/2003 or VMWare), we do have PCNS 3.0. If not then 2.2.4 is the one. Can you telnet into the NMC and perform a ping test going to the IP of the PCNS?

[BillP]

Okay, let us try PCNS 3.0 and let us know of the outcome.

[Anonymous user]

When NMC's mask is set to 255.255.255.0, it always says
57058 274.257207 10.0.1.82 10.0.1.255 UDP Source port: 40838 Destination port: apc-3052

When NMC's mask is set to 255.255.255.128, it always says
3949 19.936337 10.0.1.83 10.0.1.127 UDP Source port: 41948 Destination port: apc-3052

And there is never any packet directly from the PCNS IP to the NMC's IP, no matter if PCNS can see it or not.

This should be the problem -- it only uses the broadcast IP to communicate with PCNS.

[voidstar_apc]

>
Is it possible to configure the management card or the PowerChute to not require same netmask? Or is it possible for you to develop this option in your next update of firmware or PowerChute?
>

The Network Management Card (NMC) and PowerChute (PCNS) are not required to share the same subnet mask. The NMC sends a broadcast packet if the PCNS client is on the same subnet as the NMC, and sends unicast packets if they're on a different subnet.

So it should work, and it's not working for you. When you have a PCNS client on a different network segment from the NMC:
- What are the IP/Netmask/Gateway settings on the NMC?
- What are the IP/Netmask/Gateway settings on the PCNS client?
- Does PCNS give an error of some sort?
- What do you see on the network level via Wireshark or other packet sniffers? On the NMC side? On the PCNS side?

[Anonymous user]

Hi leizhoucn, I'm a little confused about your post as the "powechute" software you have described sounds like powerchute business editon which should not be used in conjunction with an NMC. If you are using and NMC then the correct shutdown software should be powerchute network shutdown (most likley 2.2.4). Maybe you can help elaborate as to what your exact setup is there. Hope this helps.

[Anonymous user]

Yes. You are right. I am using PCNS 2.2.4

[Anonymous user]

NMC: IP 10.0.1.81 MASK: 255.255.255.128 GATEWAY: 10.0.1.1
PCNS: IP 10.0.1.1 MASK: 255.255.255.0 GATEWAY: 10.0.1.1 -- not work;

NMC: IP 10.0.1.81 MASK: 255.255.255.0 GATEWAY: 10.0.1.1
PCNS: IP 10.0.1.1 MASK: 255.255.255.0 GATEWAY: 10.0.1.1 -- work;

NMC: IP 10.0.1.81 MASK: 255.255.255.128 GATEWAY: 10.0.1.1
PCNS: IP 10.0.1.11 MASK: 255.255.255.128 GATEWAY: 10.0.1.1 -- work;

NMC: IP 10.0.1.81 MASK: 255.255.255.0 GATEWAY: 10.0.1.1
PCNS: IP 10.0.1.11 MASK: 255.255.255.128 GATEWAY: 10.0.1.1 -- not work;

When it works, PCNSConfig.sh can register the node to the card, and return successful. Then the PCNS network interface can communicate with the NMC;
When it does not work, PCNSConfig.sh still can register the node to the card, however it waits a long time and says that it cannot communicate with the card. Then if I force configuring,
the PCNS network interface cannot communicate with the NMC.

I am testing with wireshark. Will post later.

[voidstar_apc]

Hi Lei Zhou,

>
So, the problem is:

The NMC does not care who is the listener. It just broadcast the data every 25 seconds.
And the PCNS will just listen to the ip based on the netmask of the computer it is installed.
When the NMC netmask and the computer netmask are not the same, PCNS will just hear
nothing.
>

As I mentioned earlier, it only broadcasts when the clients are on the same local segment, however it will directly send the data if they are on different local segments.

>
NMC: IP 10.0.1.81 MASK: 255.255.255.128 GATEWAY: 10.0.1.1
PCNS: IP 10.0.1.1 MASK: 255.255.255.0 GATEWAY: 10.0.1.1 -- not work;
>

Here, we see the network ID of both PCNS and the NMC are the same:
10.0.1.81 & 255.255.255.128 = 10.0.0.0
10.0.1.1 & 255.255.255.0 = 10.0.0.0
So the NMC will broadcast the information on 10.0.0.127 which you can see in your wireshark packet dump.

Each case (both working & not working) uses broadcasts from the NMC, and the NMC seems to use the right broadcast address, but PCNS does not seem to be picking it up when the broadcast address doesn't correspond with the network mask. You might try using wireshark with promiscuous mode turned off on one of the PCNS machines to see if the traffic gets to the machine. If so, I'll ask the PCNS guys why PCNS isn't processing them.

Another thing you could try is forcing the NMC to use unicast by making the broadcast domain smaller. For instance:
NMC IP: 10.0.1.2
NMC NM: 255.255.255.252
NMC GW: 10.0.1.1

edited clarified a bit

[Anonymous user]

Apparently I cannot do this, since 10.0.1.2 through 10.0.1.62 are all assigned to other purposes. Any other idea? Would it work if I put the UPS ips to 10.0.3.x? I will try this when I get a chance.

[voidstar_apc]

>
Apparently I cannot do this, since 10.0.1.2 through 10.0.1.62 are all assigned to other purposes. Any other idea? Would it work if I put the UPS ips to 10.0.3.x? I will try this when I get a chance.
>

Yes, I like that even better although the NMCs will need a corresponding gateway (10.0.3.1 possibly?) which needs to be on the same physical LAN segment.

[Anonymous user]

The information it puts on the broadcast port is like:

0000 ff ff ff ff ff ff 00 c0 b7 55 bd 1d 08 00 45 00 ........ .U....E.
0010 02 60 00 01 00 00 40 11 61 ba 0a 00 01 54 0a 00 .`....@. a....T..
0020 01 7f d6 80 0b ec 02 4c 00 00 4d 41 53 54 41 54 .......L ..MASTAT
0030 55 53 0a 3c 42 4f 44 59 3e 0a 4d 56 3d 76 31 2e US..MV=v1.
0040 37 2e 30 0a 50 43 3d 31 30 2e 30 2e 31 2e 38 34 7.0.PC=1 0.0.1.84
0050 0a 4d 41 3d 30 30 20 43 30 20 42 37 20 35 35 20 .MA=00 C 0 B7 55
0060 42 44 20 31 44 0a 44 54 3d 30 34 2f 32 37 2f 32 BD 1D.DT =04/27/2
0070 30 31 31 0a 54 4d 3d 31 39 3a 30 35 3a 30 30 0a 011.TM=1 9:05:00.
0080 53 52 3d 30 34 0a 53 55 3d 38 37 36 39 42 38 0a SR=04.SU =8769B8.
0090 52 44 3d 39 44 34 33 0a 54 41 3d 31 30 2e 30 2e RD=9D43. TA=10.0.
00a0 31 2e 31 32 37 0a 55 43 3d 31 0a 53 44 3d 30 0a 1.127.UC =1.SD=0.
00b0 53 4d 3d 32 0a 35 43 3d 32 39 2e 33 33 0a 34 46 SM=2.5C= 29.33.4F
00c0 3d 31 30 39 2e 38 35 0a 35 31 3d 30 38 0a 37 45 =109.85. 51=08.7E
00d0 3d 30 30 0a 33 45 3d 30 0a 34 43 3d 31 30 39 2e =00.3E=0 .4C=109.
00e0 38 35 0a 35 30 3d 33 30 2e 33 30 0a 34 33 3d 33 85.50=30 .30.43=3
00f0 31 2e 32 35 0a 37 38 3d 30 31 2f 30 34 2f 32 30 1.25.78= 01/04/20
0100 31 31 0a 36 41 3d 31 36 33 37 0a 34 37 3d 4f 0a 11.6A=16 37.47=O.
0110 36 36 3d 31 30 30 2e 30 30 0a 35 38 3d 4f 4b 0a 66=100.0 0.58=OK.
0120 33 39 3d 46 46 0a 32 37 3d 30 30 0a 33 38 3d 30 39=FF.27 =00.38=0
0130 30 0a 36 45 3d 4a 53 31 31 30 32 30 31 32 32 32 0.6E=JS1 10201222
0140 31 20 20 20 20 0a 36 44 3d 30 31 2f 30 34 2f 32 1 .6D =01/04/2
0150 30 31 31 0a 36 32 3d 55 50 53 20 30 31 2e 36 20 011.62=U PS 01.6
0160 2f 20 43 4f 4d 20 30 31 2e 36 20 2f 20 55 42 4c / COM 01 .6 / UBL
0170 20 30 33 2e 32 20 2f 20 43 42 4c 20 30 36 2e 30 03.2 / CBL 06.0
0180 0a 30 31 3d 53 6d 61 72 74 2d 55 50 53 20 58 01 .01=Smar t-UPS X.
0190 33 30 30 30 20 20 20 20 20 20 20 20 20 20 20 20 3000
01a0 20 20 20 20 0a 32 46 3d 37 2e 36 38 0a 34 32 3d .2F= 7.68.42=
01b0 31 33 35 2e 30 30 0a 34 36 3d 36 30 2e 30 30 0a 135.00.4 6=60.00.
01c0 34 44 3d 31 30 39 2e 38 35 0a 34 45 3d 31 30 39 4D=109.8 5.4E=109
01d0 2e 38 35 0a 34 35 3d 33 33 36 0a 37 35 3d 0a 36 .85.45=3 36.75=.6
01e0 43 3d 0a 36 35 3d 0a 36 46 3d 31 32 30 0a 37 31 C=.65=.6 F=120.71
01f0 3d 32 0a 36 42 3d 4f 0a 37 30 3d 32 30 0a 37 32 =2.6B=O. 70=20.72
0200 3d 31 30 0a 38 31 45 32 3d 0a 30 35 3d 30 30 0a =10.81E2 =.05=00.
0210 30 39 3d 36 30 2e 30 30 0a 49 30 3d 30 0a 53 48 09=60.00 .I0=0.SH
0220 3d 43 31 31 42 46 37 46 33 33 30 30 33 46 44 37 =C11BF7F 33003FD7
0230 45 43 33 46 43 39 42 33 34 43 34 32 41 46 31 30 EC3FC9B3 4C42AF10
0240 41 0a 4d 44 3d 30 62 36 36 64 66 39 39 62 64 62 A.MD=0b6 6df99bdb
0250 62 62 30 33 39 34 36 35 35 32 31 62 66 66 39 34 bb039465 521bff94
0260 66 38 33 62 62 0a 3c 2f 42 4f 44 59 3e 0a f83bb..

which read clearly the measurement of the UPS sensors.

So, the problem is:

The NMC does not care who is the listener. It just broadcast the data every 25 seconds.
And the PCNS will just listen to the ip based on the netmask of the computer it is installed.
When the NMC netmask and the computer netmask are not the same, PCNS will just hear
nothing.

[Anonymous user]

And I have just tried, if I trick PCNS by setting its netmask as 255.255.255.128, while the actual netmask of the computer is 255.255.255.0, and have the NMC to be masked for 255.255.255.128, it still does not work.

For now I am setting a third ethernet port dedicated for communicating with the NMC, and it worked. However this is a waste of resource.

It is still very appreciated if you could make the firmware directly talking with the registered IP instead of broadcasting the UPS data in the next update of the firmware/PCNS/

Thanks.

[Anonymous user]

Yes, I tested with firebind.com

[Anonymous user · Accepted solution]

It is win server 2003 and 2008. I will try 3.0