Continued from Part 3…
If you have set up an OpenVPN server on your system and are using it regularly, eventually you are going to want to trim the log file. Webmin actually makes that easy. Simply click on System, then Log File Rotation. You should see a bunch of existing log file rotation rules. Up near the top of the page there’s a line that reads:
Select all. | Invert selection. | Add a new log file to rotate.
Click on Add a new log file to rotate. You should get a page that looks like this:
The main thing here is to get the correct log file path into the topmost text area. The path will be something like:
I generally keep all the default settings except for these two:
Rotate even if log file is empty? (I set to No)
Ignore log file if missing? (I set to Yes)
But you can do as you wish. The important thing is to make sure that the log isn’t simply allowed to grow forever Set it up as you like, click Create, and you’re through.
And now a note for FreePBX and Asterisk users. When setting up an extension, if you use the permit and deny fields to enhance security, the correct way to fill these out may not be intuitive. For example, if you do sip show peers from the CLI, an extension at the client end of the tunnel may show up with an address in the range of addresses assigned by the client router (such as 192.168.5.x) and yet when you fill out the permit field, using that address may not work. Asterisk’s log file will generally tell you the address it wants to see, and in our case that was 10.8.0.10! No, I don’t know why, but just wanted to give you a “heads up” on that one.
I had mentioned in Part 3 some of the things that needed to be done if, from machines on the server side of the VPN tunnel, you wanted to be able to access machines at the client network (where the router with the Tomato firmware is located) that are on the WAN port side of the router. Bear in mind that anything connected to one of the LAN ports on the router is considered to be part of your VPN, but sometimes you might wish to access a machine or device (such as an “upstream” router) on the WAN side of the router with the Tomato firmware. To do this, you need to add the route to the WAN side network in the server configuration (in the “up” and “down-pre” script sections at the bottom of the Webmin server’s configuration page, using an additional “route add” and an additional “route delete” statement), and then on the client configuration page you must add an additional iroute statement – all of those take the same format as the lines you added to access the network on the LAN side of your client. At that point, you can access machines on the WAN port side of the Tomato router, but it’s not reciprocal – they can’t access machines on the server side.
Now, I need to make an important distinction here – I’m talking about machines connected to the WAN side of the Tomato-firmware router. Anything connected to one of that router’s LAN ports should already have full access to your network (on the server side). But the thing to remember is that ANY traffic sent out by a client connected to the LAN side will go through the tunnel. In some cases that may not be the desired behavior – you might have a few devices that should use the local Internet connection for all outgoing traffic (that is, as a rule they DON’T send their traffic through the tunnel), BUT you’d like to make an exception so that they can access only the local network on the server side of the tunnel, so that “local” traffic CAN be routed through the tunnel. So, you’d have such devices on the WAN port side of the Tomato-firmware router (that is, connected to the same “upstream” router or switch as the Tomato-firmware router) so they don’t use your tunnel for the bulk of their traffic.
So the question then becomes, is it possible to allow those devices to use your tunnel ONLY for traffic to the local network on the server side of your tunnel? Well, it is, but it’s a bit tricky to set up. Note that you MUST first have it working in the opposite direction (that is, at a machine connected to the server side of the network, you can reach machines on the WAN port side of your Tomato-firmware router – that’s what I was talking about a couple of paragraphs up). If you can’t do that, you’re not going to get it working in the opposite direction. If you CAN do that, then here are the additional steps:
In the Tomato-firmware router, click on “Advanced” (in the left-hand menu), then “Firewall”, then check the box next to “Respond to ICMP ping.” You should now be able to ping the Tomato-firmware router from another device on the WAN side of the network (which may be important for testing and troubleshooting).
Next, click on “Administration”, then “Scripts”, then click the “Firewall” tab. You should see a big text entry box with (probably) nothing in it. Enter lines similar to the following:
iptables -t nat -I PREROUTING -s 192.168.10.0/24 -d 192.168.0.0/24 -j ACCEPT
iptables -t filter -A wanin -s 192.168.10.0/24 -d 192.168.0.0/24 -j ACCEPT
iptables -t filter -A wanout -s 192.168.0.0/24 -d 192.168.10.0/24 -j ACCEPT
In this example, addresses on the WAN port side of the Tomato-firmware router are in the 192.168.10.x range, while addresses on the server-side LAN are in the 192.168.0.x range. If either is different on your system, be sure to change all three instances of the appropriate base address.
Then click the Save button at the bottom of the page. After that it should look like this:
Reboot the router (or you can ssh in and manually enter each of the lines from a command prompt, if you want to avoid the reboot). Now any traffic for the server-side LAN that reaches the Tomato-firmware router will get passed through the tunnel, but you still need to instruct the individual machines or devices to route that traffic correctly (which may be easier said than done for some machines). I don’t know how you do it from a Windows box, but I can tell you how it’s done on a temporary basis (that is, it survives until the next reboot) on a Linux-based or Mac OS X based machine. For the sake of these examples, assume the Tomato router is at (and can be pinged at) 192.168.10.50:
From a Linux box:
sudo route add -net 192.168.0.0 netmask 255.255.255.0 gw 192.168.10.50 eth0
(eth0 is the name of the interface used to connect to your local network)
From a Mac OS X box:
At a terminal prompt enter:
sudo route add -net 192.168.0.0 -netmask 255.255.255.0 192.168.10.50
Then, if there are shares on server side of the network that you want to connect to, and you know the host machine’s IP address, open a Finder window, click on “Go” in the top menu bar, and enter this as the destination (substituting the correct IP address for the target machine):
Note that in at least some cases, the connect attempt will fail if you don’t explicitly specify the port (:139) – this is apparently some kind of bug in recent versions of OS X.
If anyone knows how this is done on a Windows box, or how to make these route statements persist after a reboot (remember, they must be run by the root user or a user with root-level privileges, which is why the sudo statement is used — and you can’t put sudo in a script because it prompts for a password), please leave a comment and share your knowledge!
If you have followed this series thus far, I should point out that these articles are not static – if I find a mistake, or a better way to do things, they may get changed. On the other hand, since this particular router probably won’t be in my possession much longer, it may be something that I don’t do much more work on.
One thing I had said I would do in this last article is to give you a list of links that I found useful, or at least interesting, while working on this project. I didn’t actually utilize the information in all of these, and some are even a bit off-topic for the subject at hand, but this is just a small fraction of the pages I went through while trying to get this to work:
The ‘Point and Click’ Home VPN HowTo Guide (this was one of my primary sources)
OpenVPN: Building and Integrating Virtual Private Networks (book – Amazon affiliate link)
Tomato’s Frequently Asked Questions & Tips
Tomato (firmware) page at Wikibooks
An Easy Guide to Installing Tomato on the Asus 520gu
Teddy_bear’s Tomato 1.25 ND USB + FTP/Samba Mod (In my opinion the best firmware mod for Asus WL-520GU – be sure to get the VPN version)
Keith Moyer/SgtPepperKSU’s VPN build with Web GUI (also a great version, particularly if your router isn’t supported by the above version)
thor2002ro’s SDHC | SNMP | VPN | USB Mod (includes features from both of the above versions plus some additional features, but note that latest versions won’t run on routers with insufficient memory).
Summary of OpenVPN settings in Tomato Firmware
Setting Up A Low Cost NAS Using Tomato
Did I configure QoS VoIP correctly? (message thread)
QOS for SOHO VOIP Solved, Tomato Firmware
Optware installation instructions (supposed to also work with Tomato firmware, potentially allows use of numerous software packages originally written or converted for Linksys NSLU2)
OpenVPN IPv6 Tunnel Broker Guide
OpenVPN client configuration for Windows, Linux, Mac OS X and Windows Mobile for Pocket PC
Installing a Virtual Private Network with OpenVPN
EDIT: Create a VPN with the Raspberry Pi
USB disc partitioning utilities available.
“A set of disk utilities that will execute on a Tomato router. With these utilities you can now create ext2 partitions on a USB drive on the router itself, so you don’t have to use a Linux desktop machine to do it anymore.”
“A brief help file is included.”
Download link (filename is “tomato_dskutils.tgz”)
And there’s probably plenty of other great links that I’ve missed.
Finally, one more word about the TAP/TUN issue. I would sort have liked to have gotten this working in TAP mode. However when I tried to set it up, OpenVPN (on the server) complained about a missing brctl file. Well, it turned out that the way to get that file was to do yum install bridge-utils – sounds easy, right? I assure you, absolutely nothing about this project was easy, at least not for me.
The problem was that after I had installed the software and switched both sides from TUN to TAP, and then restarted the OpenVPN server, it brought down the entire local network! I mean to tell you, I couldn’t connect to any web pages or do anything else until I physically killed the power to the server box! When I brought it back up and disabled OpenVPN, everything connected to the LAN worked fine again. When I uninstalled bridge-utils and went back to using TUN, the tunnel started working again. I had been up all night, it was coming up on 7:00 AM, and I was just so doggone frustrated by that point that I never even tried to get TAP working again. Besides, I just don’t like doing things that can bring down the entire network. I suspect it was doing some kind of packet flood thing, sort of a denial-of-service attack on my local network – pardon me if I’m not thrilled about the prospect of trying that again!
After some additional online research, I suspect that part of the problem is that after installing bridge-utils, you need to create and/or modify certain files, such as /etc/sysconfig/network-scripts/ifcfg-br0, /etc/sysconfig/network-scripts/ifcfg-eth0, and possibly /etc/sysconfig/network-scripts/ifcfg-eth1 (though I’m not at all sure about that last one). For example, one site I went to (which, for some reason, I could only read by using Google’s cached copy, which is why I’m not giving a link) said, “Configure this server’s network configuration to use a bridge as its primary interface. You do this by bridging the physical ethX and virtual tapX interfaces into one logical br0 interface. The br0 interface will be assigned an IP address, and not the physical or virtual interfaces.” That site also suggests that those files should read as follows (note that I do not recommend following this advice verbatim, see my comments below):
IPADDR=192.168.0.50 <— local IP of the server
Please note the above is totally untested at this point, and I’m afraid that the advice to modify the existing files (particularly /etc/sysconfig/network-scripts/ifcfg-eth0) may (or may not) be ill-advised. What concerns me is that the /etc/sysconfig/network-scripts/ifcfg-eth0 file seems to contain a lot of essential information that is in effect being thrown out – for example, on our system, it reads as follows:
TYPE=Ethernet <— NOTE!! 'Ethernet', not 'ETHER'
I’m just not sure what is the proper thing to do here — maybe just add the BRIDGE=br0 line to the existing file? But, if you do decide to try a full replacement of any file, be sure to copy the existing file to a safe location so that if things go badly you can recover your original file!
Some of the comments I have read suggest that TAP mode is not as efficient in transferring data, and/or not as secure (unless you add even more configuration options), so I’m thinking maybe we should leave well enough alone. But, if you have a truly burning desire to get it going, I suggest using the following Google search string for additional information – it may be strange, but it actually produced the most relevant results of all the searches I’ve tried over the last few days:
Hopefully this page won’t show up as the first result! 🙂
If you have any ideas about what went wrong, or in particular, if you manage to get this working in TAP mode, I’d be most interested to hear about it (and how you did it, if you got it working) – the comments are open.
EDIT (November 30, 2011): While I’m not really wanting to reopen this project at this late date (this still remains about the hardest thing I’ve ever tried to do with a computer, and I have a distinct aversion to revisiting it), I did receive an e-mail today from James R, which I will post verbatim here. NOTE THAT THIS IS NOT TESTED BY ME, SO USE AT YOUR OWN RISK:
From: James R (address redacted)
Subject: OpenVPN in bridged mode
Date: November 30, 2011 11:32:40 AM EST
The following is a script to fix the problems with getting bridged mode OpenVPN working with PBX in a Flash (CentOS 5.7 with Webmin). First you install the third party Webmin OpenVPN module, then use the script below. I haven’t tested it yet, but I believe it should work if not explain what actions needed to be done to repair it. Be kind, as my scripting skills are quite poor.
rpm --import http://apt.sw.be/RPM-GPG-KEY.dag.txt
rpm -K rpmforge-release-0.5.2-2.el5.rf.*.rpm
rpm -i rpmforge-release-0.5.2-2.el5.rf.*.rpm
yum -y install bridge-utils
yum -y install tunctl
yum -y install openvpn
echo "start_cmd=/etc/init.d/openvpn start
" > /etc/webmin/openvpn/config
cat /usr/libexec/webmin/openvpn/br_scripts/bridge_start | sed
cat /usr/libexec/webmin/openvpn/br_scripts/bridge_end | sed
(End of James R’s e-mail. I fixed some punctuation and capitalization in the first paragraph, but otherwise it’s the way he sent it. I was NOT sure if the final couple of sections were really supposed to be three lines each, or one line each that got broken up by the e-mail software. I suspect the latter, but I’m leaving them as is in case I’m wrong about that. Again, please remember that the above is UNTESTED by me.)
Here’s another bit of information that may be useful for those of you that don’t know much about Linux — here is a very small list of Linux commands that may be useful in diagnosing any problems with your VPN tunnel:
- ifconfig – shows the current list of network interfaces. On both ends of your tunnel you should see a tunx interface when the tunnel is operational. On Windows-based systems a similar command is ipconfig.
- ip route show – shows current routing information for the system (see also route). ip route list gives a slightly different view.
- iptables -L – lists the current iptables rules. Add the -v option to get a more verbose display.
- netstat -r – similar to route but with a slightly different view.
- ping address – tries to get a response from another connected machine or device. Note that not all systems or devices will respond to pings.
- route – shows the current routing tables on the system (see also ip route show).
- tcpdump -n – this shows a running display of all activity on the network interfaces. Be careful because this can produce a LOT of output very quickly. Use Control-C to interrupt, then be prepared to wait until the buffer empties (may take a few seconds).
- traceroute address – If you run a traceroute to a network address (either on the LAN or on the Internet) it will attempt to show each system the packets pass through on the way to their destination. This can be useful for determining if traffic to a particular destination is actually going through your tunnel. On Windows-based systems use tracert (a holdover from MS-DOS days when filenames were limited to eight characters!).
- which program-name – not a network command per se, but if you get an error message about a missing program, you can use which program-name to try to determine if the program exists on your system, and the correct path to that program.
Note that there are additional options for most or all of the above commands – read the man page for that command (e.g. man tcpdump) if you are interested, or use a search engine to find more information (yeah, I think most man pages are painful, too). man is short for manual, by the way, not a reference to gender.
Anyway, I’m still trying to catch up on lost sleep, but if I think of anything else pertinent I’ll probably add it to this article, rather than making this series any longer. I hope if you attempt this, it’s not nearly as painful for you as it was for me!