Error on non-standard subnet mask

[aramsay]

I've just upgraded (again) from 3.1.65 to 3.2.17; all looks relatively OK, however in Manager we get a warning which says "Field 2 must be 255 to be valid netmask for specified IP address".

Our subnet mask is 255.255.252.0; if I change third octet of the subnet mask to 255, then the error goes away; however there's nothing wrong with it being 252.

Is there any way to stop this error coming up every time I open the config file, as even if you don't make any changes to it, it thinks you have so says that you need to do a reboot.

 

[kholladay]

Wizard also gave an error when using subnetting a class A address to class C for example 10.0.0.0 255.255.255.0 even though this is perfectly valid.

Just ignore it. Many of the "errors" are to be treated as informational.

Kyle Holladay
Certified: ACACN, ACSCI, ACSCM, TIA-CTP
"If it worked the way it should you wouldn't need me

 

[aramsay]

Yes, it is possible to ignore; but it's a PITA as even although you don't change the subnet, everytime you go to send the config it says that it's had a change to a non-mergable item, so wants a reboot.

 

[siftach]

If you don't use your IPO as a router, why don't you just give it what it wants? Give it an address that will comply to a C-Class subnet with your router and it will stop giving you that error.

 

[aramsay]

That would then stop all the clients from being able to see the IPO for Phone Manager, Soft Console; stop the IPO being able to see the VMPro server, etc.

Unfortunately that would not go down well at all!

 

[Spanning]

the ipo does not like subnets , have YOU tried creating a static IP ROUTE to your NETWORK gateway address , this will then allow your other users to see the IPO

 

[siftach]

Like Spanning says, a static IP route will solve the problem if it happens. As long as IPO can reach the router, the router will take care of routing the traffic.

 

[aramsay]

OK, now I'm totally lost.

Our network has IP address 192.168.64.0/255.255.252.0 with our router being 192.168.64.1, the IPO is 192.168.64.5

What it appears that you are suggesting is that the IPO is configured as 192.168.64.5/255.255.255.0 with a static route to 192.168.65.0/255.255.255.0, 192.168.66.0/255.255.255.0 & 192.168.67.0/255.255.255.0 all going to 192.168.64.1

Whilst this will allow the IPO to be able to send data to the client machines (via the router (and putting extra load on it too)), however the client machines would try to send data back directly to the IPO as they are on the same subnet; however the data would never get to it.

 

[siftach]

You have a real problem with your non standard configuration: You are using C Class addresses on a network bigger than a C class. I would try to do two things:

1. Set the IPO netmask to 255.255.0.0 - after all, any address starting with 192.168 will be local. However, since you will be using a C class address with a B class netmask, I'm afraid IPO will still have the same problem.

2. If 1 doesn't work, try creating a totally different netmask for IPO, something like 192.168.1.x, and configure your reouter to support both networks.

3. What you really should do is set your network addresses right. Either work as a full B class (255.255.0.0 netmask) or setup 4 difference C classes (192.168.64.x, 192.168.65.x, 192.168.66.x and 192.168.67.x). But that will be allot of work, I'd just go and kicj the guy who planned this network in the ***.

 

[aramsay]

Since the introduction of CIDR (

http://en.wikipedia.org/wiki/Classless_Inter-Domain_Routing)

in September 1993, it has been standard practice to use "supernetting" to combine multiple contiguous blocks into one network, in a similar way that subnetting is used to split one contiguous "Class C" IP address range into multiple ranges. We all need to forget about the old classful ways, which meant that there was only Class A, B, C, D and E (although I'm not sure that Class E was actually ever used).

If every company that has an internet connection and had been allocated a handful of IP addresses (lets say a /29 in CIDR notation (255.255.255.248 (ie 6))) had been allocated a full "class C" ip range, then we would have needed IPV6 many years ago.

It is exactly the same for our networks "Supernet" (

http://en.wikipedia.org/wiki/Supernet)

range (/22 in CIDR notation (255.255.252.0 (ie 1022 IP addresses))) is perfectly valid.

 

[siftach]

I guess you are right, I forgot how bad Avaya's IP and lan equipment is.

 

[aramsay]

I really didn't feel like kicking myself!

 

[siftach]

I just put your subnet on my testing system, and Manager didn't require a reboot. Maybe it's because I didn't save the changes as it requested (this is probably a bug in manager - no changes were made and still it wanted to save it), but anyway I don't reboot the systems everytime Manager thinks it should.

This is an annoying bug - everytime you open Manager this error, which is actually not described as an error, will pop up and if you move away from the system tab it will ask you to save and then ask you to reboot if you did save. You should write about this to Avaya, so in the next release they will fix the bug.