So yesterday, Thursday 10/30/2008, I'm having a fairly normal day. I'm not feeling so well so I don't plan any exercise today despite the beautiful weather in Laramie.
Around 6 PM I decide to check up on my web sites and download my email as I do periodically throughout the day, and I find that I cannot get to my main web site www.plainsboro.com, nor can I download my email or log into the site via ssh. After checking a few things out (yes, I can get to other sites just fine, etc.), I give my web server ISP--Jumpline--a call. The response: nope, your web server is up and running. OK; fine. Some glitch with my home ISP, Bresnan Communications Cable service, then. Such glitches are not all that uncommon in the connect-to-the-internet-via-cable-modem world. The night wears on and still no connectivity to my web servers so I sleep on it.
The next day (Halloween!), the same thing--I find that I can't get to my web site in any fashion. I also discover that I can't get to Jumpline's tech support web site either, plus several other web sites they host (like www.midwestclosets.com). Hmmmmm. I have one other thought at this point, which is to use a proxy host and try to connect to my web site. In general, I hate using proxy web hosts because you have no idea what they might be doing to you: capturing all your keystrokes trying to find passwords, etc., plus they're often located in foreign coutries like Iran or China. But I eventually find a free public web proxy in the USA where I can just type in a URL. They basically let you do this for free, but they wrap your browsing with advertisements. As it turns out, this works for me. I can connect to my web server via a proxy host just fine. But of course, this doesn't solve my ssh or email needs, and I definitely don't want to use a proxy host for those connections.
But I'm able to get to Bresnan's tech web site and I'm able to chat online with a network support person there. The basic answer I get is that you're either connected to the internet 100% or disconnected 100%, and if you are having problems then it has to do with your computer's configuration or a firewall that is blocking certain sites. Not too helpful. Now, I've been a Unix and TCP/IP network manager for about 24 years and this is definitely a frustrating problem, but I know damn well that I'm not running a filtering firewall that is keeping me from getting to my own web server.
However, I do have a couple thoughts: (1) several months ago, Windows Vista's automatic update ran on my computer and I found that on the first boot after its update install, the internet was broken. At that time I checked my connectivity with my backup computer, a laptop running XP and that computer's connection to the internet seemed OK. So I just rebooted my Vista box again and that seemed to solve the problem. This is amazing, isn't it? Over 20 years ago when Windows first came on the scene, their solution to every problem was to reboot your computer. Here we are in 2008 and rebooting the computer still seems to be the #1 way to fix a problem. Yikes. (2) The Bresnan support technician suggested I try a second computer and to also try bypassing my cablemodem router, both good suggestions.
So I try both of these things. Rebooting Vista doesn't solve the problem. And I gave away my backup computer running XP so I dig deep and try to use a *very* old laptop, a 450 Mhz Acer I bought in 2000. It still runs, but it freezes all the time and isn't too useful for much these days. I fish it out of storage and hook it up in dhcp mode; it manages to connect to my local network and to the internet, but it has the same symptoms as my Vista box--no connectivity to my own web sites. I try the second thing--bypassing my cablemodem router. Luckily, Vista is pretty savvy about dhcp and I don't need to reboot it when I plug my ethernet cable directy into my cable modem instead of my cablemodem router. It just figures out it's on a new network and refreshes dhcp. Interestingly to me, I find that I am assigned a *different* IP address than the external IP of the cablemodem router gets assigned, but as I think about it, it makes sense since the dhcp lease on the cablemodem router is probably still valid, and might even be tied to its MAC address. Alas, bypassing the cablemodem router doesn't fix the problem either.
OK; I also reboot all of my networking equipment--the cablemodem router, my cable modem, and even the little 100Mb switch I use. Interestingly, NONE of these pieces of equipment has an on/off switch. To reboot each of them, I have to unplug their electrical sockets, wait 10 seconds and plug them in again. All of the equipment seems to be functioning normally.
But I still can't get to my web sites, dammit!
So I'm very frustrated by now, and even though I'm a bit under the weather, it's another beautiful day here, mostly sunny with temps around 67 degrees, so I decide to run an errand on my recumbent trike and also take a long-ish ride on the Laramie Green Belt. I end up going about 13.5 miles in 82 minutes with a stop at the Post Office, and during my ride I remember that I also have an iMac computer that I hardly use but which can now become my backup computer. So when I get back from my bike ride, I'm a bit less frustrated and I have another thing to try. I boot the iMac, plug it into my local network and also try bypassing my router, but once again even the iMac can't get to my web sites.
OK; so I've tried three different computers, three different operating systems, both attached to my router and bypassing it. Six configurations total, and all fail. I think about this, then decide that since all the sites I can't get to are at Jumpline, I'll try their customer support line again. They blew me off yesterday, but maybe there's some new information.
Sure enough, there's now a message on their main support hotline explaining that there is a routing issue between Sprint and Cogent. And in fact when I do a search in google news, I find that it's more than just a routing issue--these two internet backbone companies have decided to completely stop talking to each other. Yikes. For more information about this, do a google news search for "Sprint Cogent" and look at the news article results for October 30 and 31 2008.
Now, I know a bit about TCP/IP. It's a very robust set of protocols with sophisticated routing protocols on top of it. When there's a problem, it's generally smart enough to automatically re-route traffic so poor old users like myself don't even notice a problem. Remember 9/11? The World Trade Center towers supported much wireless and wired infrastructure critical to internet and cell phone operations. When the towers fell, lots of cell phone service dropped out, but the internet itself largely stayed up and became an important communications tool during that time period.
Not so today, apparently. It seems that today because Bresnan Communications connects primarily to Sprint and Jumpline connects primarily to Cogent, because these two companies are currently refusing to exchange backbone internet traffic with each other, little old me sitting in Laramie, Wyoming now cannot connect to my web servers. Argggggghhhhhh.
This is truly horse hockey, bull crap, or whatever nastiness you want to call it. Large companies like Bresnan and Jumpline should have backup plans for this type of problem. At the user level where I'm at, I shouldn't have to worry about issues like this myself. What do other folks who don't have internet administration experience do? Do they just say: "oh well, I guess these sites that I usually visit are down and I can't do anything about it"? Apparently so. And with the initial customer support that I got from both Jumpline and Bresnan, the basic reaction by both customer support folks I communicated with was: "Everything looks fine from here; this must be your problem or at least it's certainly beyond our control". That's BS, too. The management teams at both companies should become aware quickly about problems like this and should inform their staff about how to advise their customers. I mean, I actually had to inform Bresnan that this problem existed and might be affecting some significant portion of their customer base. Good grief golly gee.
So now we're at about 35 hours of no Sprint<-->Cogent communication and nobody seems to be doing anything about it except filing lawsuits. And in fact since all of this took place on a Thursday and wasn't fixed by Friday afternoon, I almost certainly can expect the problem to linger at least through the weekend. Great; so next week some time I'll have 9000 spam emails to download to get my 11 actual meaningful emails.
I don't give a crap whose problem it is, who caused it, what contract negotiations are taking place, etc. SOLVE THE PROBLEM Sprint and Cogent. Or better yet, GO AROUND THE PROBLEM, Bresnan and Jumpline! SHEESH!
I'll most certainly post updates as this fiasco continues.
--SDP
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment