Resolving servers - wrong IP addresses shown.

[sv]

Distributed rendering has NEVER EVER worked for me, which is really starting to f**k me off - not in almost two years since 1.46 have i ever got a DR render bucket working. I've endlessly trailed through the forum and emailed Chaos, but to no availe.

One thing i'm noticing now, is that when i try and resolve the serves, the wrong IP address is displayed - and it's the same for each server, for instance:

sv001 IP: 217.77.179.100
sv004 IP: 217.77.179.100
etc.

but it should be:

sv001 IP: 10.0.5.4
sv004 IP: 10.0.5.19
etc.

Warning messages are displayed which say:
warning: Render host svse(217.77.179.100) is not responding.

the host computer's name where the license is stored is SVSE, not the servers where vrayspawner60.exe is running.

We have about 16 rendering machines, that are absoluetly fine for backburner - never had any problems. Possible reasons for this to fail i have tried to list below:

1) The Vray License is stored on a work machine running XP Pro, rather than the server running Server 2003 which holds all the maps and rendered frames.
2) The work machine’s license is stored on a D: drive (D:\Program Files\Common Files\ChaosGroup\VRFLServer.ini), not C: where ever other machine has Vray installed to a C: drive.
3) All machine’s are part of one network, but there are different workgroups – some machines are part of a domain, other’s aren’t. – this difference in workgroups wasn’t seen to be an issue as all machines will render properly through backburner (16 machines) and all machines can access each other through the network.
4) All 3dsMax.ini files on all computers are exactly the same, except the .ini file stored on the work machine which hold’s the license, where all C: extensions are replaced with D: extensions.
5) We have no internal firewalls, there are no permissions to stop anything through the network.


All servers with the Vrayspawner60.exe file will start, and Max will sit minimised in the taskbar. If I enter the I.P address to find the server – 10.0.5.19, then this address is resolved, but DR doesn’t work. I get the warning message:
warning: Render host svse(10.0.5.4) is not responding

Again, this is the wrong computer name for that IP address.



Anyone? Please - i will love you forever.....

[mike.edel]

afaik, in subnetted networks access via IP without knowledge of the netmask is impossible (because you are using subnets you need to know in which subnet the ip is located in)

you either need to specify both ip and netmask (if possible) or put all DR computers in one subnet

connecting by machine name works, because windows translates the name into ip+netmask (either via a DNS or with the .hosts file)

btw: tcp/ip broadcasts also don't cross subnet borders so if DR tries to find the slaves via broadcasted packets, it fails

you might want to look at
http://en.wikipedia.org/wiki/Netmask
and
http://compnetworking.about.com/od/w.../aa043000b.htm
for clarification

all in all, i wouldn't blame vray in this case - you might want to try putting one machine into the same subnet as your render client (your workstation)

If you are not able to access the machines by their internal ip address DR wont work in any case.

ps: your local ip adresses look strange to me - are you sure they have to be Class A ips?
usually the structure is something like:
internal fixed ip range = Class C (192.168.0.0 - 192.168.255.255)
-->router-->external (fixed or dynamic) ip adress depending on the provider's ip space (typically Class A (1.0.0.0 - 126.0.0.0))

[sv]

you might want to try putting one machine into the same subnet as your render client (your workstation)

All computers are within the same subnet - 255.255.255.0
I've just checked them all.

If you are not able to access the machines by their internal ip address DR wont work in any case.

I can access all machines via their internal ip address through windows explorer - i.e. just typing \\10.0.5.19 and the machines drives and shared printers etc are displayed.

Saying this, i'm far from a network specialist, and our IT guy is away for a few days. I'll have an indepth read through those web sites and see if i can see anything. But basically, we are all on the same subnet.

[SteveC]

Hi sv, I agree with mike, with regards to using machine names, messing about with ips and subnet masks is way too complicated for me, i just typed in the computers names and let windows sort out the ip stuff, i've not had a single problem doing it that way.

Only thing i can add to the name using thing is that you should'nt put a number in front of your machines name or give a number for your computers name such as 01 or 01slave, cant remember why, someone else may be able to shine some light on this, not sure if its a backburner thing.

Also like mike said, your ips look strange, when i typed in my computers name and did the resolve servers thing, my ips came up as 192.168.0.1 and so on.

Try naming em, it should work.

steve

[sv]

cheers steve; like you said messing around with ip's and subnets is way over my head aswell, but i have typed in the computer's names and they come up with the wrong IP's - glad you haven't had a problem, but this is exactly my probelm - resolved IP's aren't right and DR doesn't work as a result.

All my PC's are also named SV001, not 001SV for instance.
None of the 16 machines have numbers at there beginning.

Try naming em, it should work.

If only mate....

[SteveC]

I feel for you mate, what is your set up?

I recognise the ip 10.0.0.1, i have to type this in to access my broadband modem.

To access the switch which looks after my network i have to type in 192.168.0.1, which i take it is where all my network ips come from.

Something else which you could try, cant remember where i got this info from either, is to run backburner server once on each of the machines that you want to connect to DR.

When i use to use final render i had a problem with render slaves not being found, i was told to run backburner once on each machine that i wanted to DR to, beleive it or not DR worked after that.

Good luck mate hope you can solve it soon.

steve

[sv]

run backburner server once on each of the machines that you want to connect to DR.

We run backburner almost every night, and this has worked fine for years. I think i'm getting somewhere with mike.edel's advice because i think the default subnet mask for a 10.0 etc. network is 255.0.0.0 not 255.255.255.0. It might be a simple case of changing all computers in our office (maybe 20 - 30) and our Nas and router to be on this subnet, but obviously this is no small task and i need to speak to our network guy before doing this.

I have changed my subnet to 255.0.0.0 and one render node's to the same, and believe it or not, the DR resolved address was found correctly as 10.0.5.17! - But DR still didn't work! This is one step closer though, and i'll show and talk through it with our network guy when he's back.

Everything on our network, including backburner, has worked properly for ages, and i'm a firm believer in if it's not broken, don't fix it - but now something doesn't work and it does need to be fixed...

[mike.edel]

oh, i misconceived your workgroups with subnets
are you sure on the ip addresses?
you can check them by locally executing ipconfig /all from the command line

windows-key+R
cmd enter

ipconfig /all

you need to do this locally on all affected machines
make sure the subnetmasks are identical and the ips are in a legal range (i still regard Class A ips as a strange thing for internal networks)
if i remmember correctly, windows sometimes has problems finding computers on different workgroups - ips should be safe as long as they are in the same subnet

you should a least be able to ping different computers in the same subnet using only their ip - this is similar to what the DR client does (sending tcp packets)

you could also try watching the network traffic using tcpview (freeware)
http://www.sysinternals.com/Utilities/TcpView.html

[sv]

mike.edel,

"talking" to pc's is not a problem - i can ping all machines, see all machines through windows, they all work and have done for years through backburner, i can navigate to all machines, copy files from them, write to them, map network drives etc. etc. etc. etc.

The ONLY thing that doesn't work on our network is DR.
I feel like i'm one step closer by changing the subnet mask to 255.0.0.0 because then Vray will resolve the correct IP addresses of the machines, but DR still didn't work. Noted that i only changed the subnet mask on two machines, so i definietly need to have a chat with our network guy when he's next in.

[mike.edel]

normally you have an ip range that is used inside your local network
typically you use a class C net for this (the submask usually is 255.255.255.0 and the ip range starts with 192.168.0.0)
others are possible, too

this ip range is reserved for internal networks --> it can not conflict with ip adresses issued by providers
quote from wikipedia:

Private subnets:

The Internet Assigned Numbers Authority (IANA) has reserved the following three blocks of the IP address space for private networks:
Network address range CIDR notation
10.0.0.0 - 10.255.255.255
172.16.0.0 - 172.31.255.255
192.168.0.0 - 192.168.255.255

so the IPs you use are part of a private network and should be fine - im just not used to those as the places i've been at have always used cass c private ips (192.168.)

the subnet mask for IPs of the range 10.0.0.0 - 10.255.255.255 should be 255.0.0.0
make sure all computers that belong to the same subnet (probably all) have the same mask (otherwise you are in fact installing a new subnet)

are you running the spawner by hand or as a service?

ps: edel is my surname, i'm fine with mike

[sv]

mike,

spawner is being run by hand, not as a service.

All computers do belong to the same subnet, but it's 255.255.255.0 not the default 255.0.0.0 for a 10.0.5.0 network. This is perhaps causing the problem, because like you say, maybe this is installing a new subnet and the reason Vray won't resolve the server's IP's. However, all servers and hosts are on the same subnet....

ps: edel is my surname, i'm fine with mike

I figured this, P.S. thanks for your help, i think you are pointing us in the right direction........

[mike.edel]

255.255.255.0 is wrong for sure
even if all your computers use this one, the actual network adress is calculated wrong (there is a bit of math involved how it is calculated but it involves the ip and the subnet mask - if the mask doesnt fit the ip range, strange things can happen)

found something on the math behind

Applying a subnet mask to an IP address allows you to identify the network and node parts of the address. The network bits are represented by the 1s in the mask, and the node bits are represented by the 0s. Performing a bitwise logical AND operation between the IP address and the subnet mask results in the Network Address or Number.
For example, using our test IP address and the default Class B subnet mask, we get:

10001100.10110011.11110000.11001000 | 140.179.240.200 Class B IP Address
11111111.11111111.00000000.00000000 | 255.255.000.000 Default Class B Subnet Mask
--------------------------------------------------------
10001100.10110011.00000000.00000000 | 140.179.000.000 Network Address

if you combine a class A IP with a class C subnet mask, the output network address is wrong:
your ip:
10.0.5.0
equals these binary numbers:
00001010.00000000.00000101.00000000 (10 = binary 1010; 5 = 101; all fields are 8 bit/digits long and filled up with 0)
the (wrong) subnet mask is:
255.255.255.0 = 11111111.11111111.11111111.00000000

now if you combine this (binary AND means the result of a digit is only 1 if both input digits are 1)

00001010.00000000.00000101.00000000
11111111.11111111.11111111.00000000
---------------------------------------------------------
00001010.00000000.00000101.00000000 (=10.0.5.0)

now if you combine 10.0.5.0 with 255.0.0.0 (the correct subnet mask) you get this
00001010.00000000.00000101.00000000
11111111.00000000.00000000.00000000
---------------------------------------------------------
00001010.00000000.00000000.00000000 (=10.0.0.0)

which is a different network address
makes me wonder why it has worked so far

[DaForce]

mike.edel:
Networking Nerd!!!

j/k

mmm can you talk in binary?


01110111 01101001 01110100 01101000 01101111 01110101 01110100 00100000 01110101 01110011 01101001 01101110 01100111 00100000 01100001 00100000 01100100 01100001 01101101 01101110 00100000 01100011 01101111 01101110 01110110 01100101 01110010 01110100 01100101 01110010 00100001 00100001

[mike.edel]

hehe

not really
as long as you cant write programs in machine code everything is fine i guess...

i usually do the translation with the windows calculator in scientific mode
but during studies i also had to do it on paper (hated it btw)

there are worse things - take linux umasks (settings for user rights) for example
those are octadecimals and are set up using binaries *shudder*
i always have to think really hard about them