Build not working

Currently 2:30AM. I will get back to any responses in the morning.

The problem persists.

I will download the Process Explorer and see what it says, I will try the command prompt again however I doubt it will provide what is using port 8008; since I checked it last night.

Also the thread has been set to Resolved again.

Hi Andrew, here is my netstat file. I will get the Process Explorer details in 30minutes.

link text

Alright, I have done the Process Explorer search, nothing in the list with 8008.

Hi, please could you send us a copy of the output from running the following command (the file C:\netstat.log):

netstat -ban > C:\netstat.log

You can run this in an standard windows command prompt (obtainable by pressing Windows+R, then typing “cmd” and hitting “Run” and typing the command into the command dialog).

Also, please could you download “Process Explorer” from this link, and once running, choose Find->Find Handle or DLL. Type “\Device\NamedPipe” and hit search. Order the view by Name, and see if there is anything in the list that says “\Device\NamedPipe\127.0.0.1:8008”.

These 2 communication channels are what Swarm is attempting to create when it outputs the error you’re seeing, but we’re unable to tell which one. Hopefully one of these outputs will help us narrow it down. Thanks.

Thanks for trying that - that certainly is weird. Are you starting swarm manually to get this behaviour? This is encouraging in as much as it’s obviously managing to open the named pipe at some point, which tells us the problem may be elsewhere. I can’t see why the process wouldn’t exist though. Does the process with a PID 5772 exist in the bog-standard windows task manager?

Mine is showing up 3 times:

8570-namedpipe.png

I have done the search again, this time With SwarmAgent running.
This shows up as a Non-Existent Process. When SwarmAgent is closed, it is removed from the search.

8583-while+running+swarm+agent.jpg

It is showing up as is in my picture when run standalone & when run via Build.
There is nothing running on PID 5772, besides the Non-Existent Process known as SwarmAgent.

Also, the thread has been set to Resolved again.

Oh alright, thanks for letting me know.

The (non-existent process) shown in my picture is only from the Process Explorer search.

However, in the Task Manager it shows as SwarmAgent.

Now this is strange, I have just run SwarmAgent again in standalone mode and it is showing as (SwarmAgent.exe 1232 File \Device\NamedPipe\127.0.0.1:8008) - inside the Process Explorer Search. So for some reason it is showing as its proper tag instead of Non-Existent Process.

However it is still not present in the Listening Ports.

I ran the engine while SwarmAgent was still visible in the Explorer and it did not change, I have run it once more, this time by building first and it still remains as SwarmAgent, no longer the Non-Existent Process.

Very Strange.

Alright I have turned both OpenVPN & Firebird SQL off, no rejoice.

I have snapped a screenshot of the Process Explorer when SwarmAgent is running, some task appears Red, then Green & disappears, repeats.

Ignore the resolved state - it gets automatically resolved when I comment, and automatically unresolved when you comment!

Sorry this is taking some digging. So after you’ve started SwarmAgent.exe, it shows up in the named pipe list (but under a non-existent process). Is swarm still running after that? If so, presumably it has a different PID than was displayed in ProcExp? It’s confusing that there seems to be 2 processes at play here :confused:

Ok, this feels like we’re getting somewhere. Let’s ignore the non-existent process shenanigans for now, and assume that it’s managing to open the named pipe correctly as shown in prc exp. This leaves us with only one conclusion - that the TCP channel is the one that is failing. As to why this could be the case, I’m still none the wiser.

I notice there are a few things on the netstat which although aren’t conflicting with the port assignment, could cause problems: fbserver (firebird SQL?) and OpenVPN. Could you try disabling these for the time-being? Also, what does your network connection setup look like on your PC? Do you have multiple network adapters, driver level virtualization, or anything disabled? I’m trying to think of anything that could cause the loopback connection to fail!

Great, thanks for that screenshot - it helps a lot. I’m going to have a look at this in detail as I think I have enough information now, the red rows there indicate that the process has actually managed to obtain a handle to the TCP socket. Do you get a lot of NetBT_tcpip events in there that keep going green then red, or do they just disappear when they turn red, then never show again?

Apologies for the late reply.

There are only 2 (NetBT_Tcpip) events which pop up green, then turn red followed by disappearing, once gone they come back either green, or without a highlight, then back to red and the cycle repeats.

I have checked the log, it still prints the same message.

Starting up SwarmAgent …
… registering SwarmAgent with remoting service
… registering SwarmAgent network channels
[ERROR] Channel already registered, suggesting another SwarmAgent or client is running.
[ERROR] If you feel this is in error, check your running process list for additional copies of
[ERROR] SwarmAgent or UnrealLightmass (or other client) and consider killing them.
[ERROR] Sleeping for a few seconds and trying again…

This is very confusing then, as this is exactly the same cycle as my (working) swarmagent. The TCP connection activity can only occur in this way after the connection has been established. I tried running a web server on 8008 and the swarm connection doesn’t start in any way until I close the web server down.

It appears from process explorer that your swarm agent is past the initial connection loop - I know it sounds daft, but can you just double check the agent log for the agent that is exhibiting the TCP connection traffic, and check that it is still failing to open a channel?

Yes that is fine, if you have TeamViewer, that would make the process easier since I am unable to use call at this time.

I think we might have to try debugging this on your machine then as we don’t seem to be getting anywhere. Perhaps add me to skype (Andrew.Rodham) and I can talk you through how we might be able to resolve this?
Thanks

Solved by the brilliant Andrew Rodham! Thank you so much.