Last weekend, much to my dismay, cable access to my house was rudely interrupted, causing occasional disruptions to the digital channels and completely disconnecting me from the net. Two service calls, countless reconnections, one exchanged splitter, one new cable modem, a few replaced coaxle cables, four days, and a repair to the gound-line later, our digital channel returned to normal and I was back on the web.
All seemed well again (except for one oddity which I will discuss at the end) untill I booted up my linux box a few days later: it refuses to connect to the internet.
After much googling, irc questioning, and eventually packet sniffing, I've discovered that although requests were being sent to the local Shaw DHCP server, the responces were coming back either mutated enough to be ignored (different transaction ID) or utterly broken (significantly smaller packet size) and rarely routed properly (sender ID is of an intermediate router, not the DHCP server). Now, the odd thing is that the Windows machine I am typing on has no problem receiving its IP, and neither does my mother's laptop, both connected to the same hub as the problem machine. Also, no amount of cable switching helped, including bypassing the hub completely and hooking up only the linux machine to the cable modem.
I have not called Shaw's technical support regarding the issue since I don't think I can get any useful information out of them without spending a great deal of effort and time, and it is explicitly stated on their support webpage that they will only offer help for issues regarding the "primary" machine.
So, to those who are knowledgeable about these things, what are the possible causes, and solutions, to this problem? Could my network card be dying? Have I offended the linux gods? Or is this a sign that the grand scheme of my local cable privider to take over the world has been set in motion?
On a somewhat related note, ever since my connection was restored, my cable modem has indicated constant traffic, both in- and out-bound. During packet sniffing, I found out that all that extra traffic were
ARP packets. This has never happened before, and it hasn't stopped since. On the plus side, it doesn't seem to affect my connection at all, but I rekon I will be calling support to see if resetting my connection from their side would put a stop to the waterfall of pointless packets.
I am not sure. I know ARP tables are a key in getting a good connection. When I worked for an internet helpesk and people were unable to get a DHCP resolve we would send out a request to flush the ARP tables. That solved the problem instantly.
What puzzles me is that your other computers still seem to be working fine. Could it be they had their IP put in manually? I also wonder about your home setup. What does it look like?
Did you change your MAC address on your linux box? It's quite a simple thing to do in most linux flavors. If you did, and you grabbed something else that was already out there, that could explain most of what is going on. Most switchs aren't going to be happy with two different systems having the same MAC addy, and traffic routing will be messed up. This would explain why your DHCP requests are failing and why there is all the extra ARP traffic.
There could be othere reasons, but this is just the first thing off the top of my head.
I have to suffer constant strangeness from my Cable broadband provider and I used to have a machine dedicated as a Proxy Server. But, the strangeness all went away when I installed a simple
Linksys Router to isolate my internal network. The router can be configured to access Shaw's DHCP server to retrieve its network configuration, and provide a consistent topology for your internal network. I have mine as fixed IP addresses, but you could also use the router as an internal DHCP server.
Out of curiosity, I googled for my ARP problem and came up with
this. It seems like I won't be calling Shaw support about it afterall.
I've tried manually releasing the IP then requesting for another one without a hitch on this machine. The same IP came back, but that's expected DHCP behaviour, since clients will try to retain the same IP for as long as possible. The Windows machines do not have their IP's specified manually, since Shaw does not allow static IP registrations for non-business users, and using anything other than DHCP would result in no connections at all.
The three computers in the house are all connected to a cheapo Netgear hub (EN 104, says the little blue box) via ethernet cables, with the cable modem connected to the hub's uplink port, with uplinking enabled. This set up has served well ever since it came to be several years ago. And, as coincidence would have it, all three machines have RealTek 8139 based ethernet adapters. They're also all in the same room.
As for my MAC address, the linux machine has not been tampered with during the downtime at all. Infact, it had stayed off for a day or two at least before the shortage, when I had no problems accessing school servers for doing assignments, and was not booted up again after my connection was restored.
Which Linux distribution are you using, especially: which DHCP client ?
Maybe Shaw has just misconfigured their DHCP server or their routers in a way that is tolerated by Windows clients, but not by your Linux DHCP client ?
Two ideas that might help in locating the problem:
1) Boot a
Knoppix and check if that is DHCPed correctly.
2) Connect your 'problem machine' to a DHCP server to which you have log file access and look at the DHCP server's logfiles. I think ISC's DHCPD will usually log any incoming and outgoing DHCP packets to syslog. At least, it does that on my Debian machines. Maybe that can give you a better idea of what exactly is going wrong.
Edit: One more thing: Since all your network cards are 8139 compatible, changing your problem machine's NIC should not be much effort. I would try that after 1), but before 2).
Concerning Linksys routers:
I heard Linksys was good, and got a home router of theirs. It routinely caused EXTREME latency, I'm talking up to 10 full seconds. I got two other home routers while trying to isolate the problem, and both of them were bunk too, I eventually found out. Having 3 bad routers really confused me while I was trying to isolate the problem, because everything pointed to the router being bad, but I couldn't believe 3 of them were bad. They were.
Later I heard Linksys used to be good when they were doing equipment for businesses, but screwed up big time when they went to the home router market. I don't know if they've gotten their act together yet or not, but I've learned you damn well better check reviews on the brand/model you want to get.
I've had three LinkSys routers, two being ok, and one being very poor. I've since purchased two D-Link routers, one being good, and one being very good. I'm currently using the 'very good' one.
My experience with LinkSys is like this:
- DMZ doesn't work as it should, and if you require DMZ capability, you're going to run into connectivity problems, or partial connections, or both, depending on your firmware.
- Upgrading your firmware is easy, but finding a stable version is quite the task. I rued the day I upgraded from my stable version, looking for something MORE stable... (actually, it was a security issue that forced me to upgrade).
- Latency issues. Two of the three I've dealt with weren't as good as they should have been, speed wise. The third stopped functioning all together. Using the modem directly was much faster than using it through the router.
- Customer service was pretty good.
- Port Forwarding and Filtering isn't as flexible as it should be (D-Link is better in this regard).
My third LinkSys router was a 1-port, and I tried to upgrade the firmware one last time, and it crapped on me. It's done, kaputsky, finito. It just sits there, blinking at me, taunting me, saying, "you'll never get a reliable internet connection out of me!". Thankfully, I picked up a D-Link router, one for me, and one for my brother. This little puppy has been functioning properly without incident for a good while now, and it was, for the most part, easily configurable. I routed ports and set filters etc, without too much hassle. I liked the the LinkSys interface better, but my new D-Link's hardware is clearly superior.
Everytime I read posts like that, I cannot keep myself from recommending
fli4l. It absolutely rocks, believe me :)
My ground-line mysteriously died on me again on Sunday. It's since been restored, but my DHCP problems hasn't gone away.
The problem machine is running Gentoo, using the latest version of dhcpcd. I've not tried booting with Knoppix, but I suppose that's a possibility. Perhaps pump will work better than dhcpcd. Lets hope that Knoppix's able to start KDE at all, what with the low RAM and slow processor, not to mention having to use a special boot floppy to enable CD-rom booting.
I don't have a DHCP server set up on any machines, so option number two isn't possible.
Swapping NIC's would be the last thing I would try, since hardware swapping means possibly having both machines down during the operation, however long it would take for a defenitive conclusion, and it's painful opening up my Linux box. (Although I suppose I should anyway for clean-up and maintenance)
I've installed Cygwin/X and Vim on my Windows2k machine, so at least I'm not restricted to doing assignments only from school. When I get the time, I will try my luck (however much of that I have, considering how the last month has gone) again.
PS: I just wanted to also say that I'm really, really impressed with Shaw's service. When my connection went down on Sunday, they were able to send a technician over the next morning, who confirmed that it was the same problem as the previous house call, just as I had suggested, and had someone else over to fix the landline again the same day, while I was at school. Bravo!
I've been using a Netgear, I believe the model is MR814. It has served me well without any problems. It's easy to configure it too.