Gremlins: 2 September 13, 2007

The company moves to a new location and in the process, acquires a new ISP. Mail settings are changed so that the outgoing mail (SMTP) goes through the ISP servers. Two months later, no one can send mail. The error message revealed by Outlook Express was Error 0×800CCC0F

ISP’s are wonderful people. And tech support is an awful task when you do not have access to a computer. The list of offenses are as follows:

1. The anti-virus checker is blocking the SMTP
2. The firewall is blocking the SMTP
3. The router is blocking the SMTP
4. Your incoming mail server has a conflict with the outgoing mail server

To support the first argument, we were referred to Google where a search of our error message revealed an article from 2004 in which an anti-virus program was blocking the SMTP server. Hopefully, there have been a few updates since then.

Why would anybody in their right mind turn off the virus checker? Even if the virus checker is blocking things, surely the solution lies elsewhere. Ditto for the firewall. Change the settings, but please do not turn off the firewall.

In their defense, the tech support people are attempting to pinpoint the source of the problem. Turning off the virus checker, then the firewall allows them to eliminate possible areas of conflict. My issue is that tech support seems to gravitate to these tow elements of a computer all too readily without looking at other possible sources.

Keep in mind that this is a business, and that the ISP has provided a business DSL service as opposed to a home service. This should be a technically superior service, coupled with technical support that understands, or is at least sensitive to the needs of the business customer.

Below is a flowchart outlining the list of suggestions made by the technical support rep I spent 1.5 hours with. This is only a partial list, but I thought I would include only the parts that had positive outcomes.

The ISP does not provide support for router? What?? On a business account? I am not sure why the tech support rep would even ask if the computers were connected through a router. I did mention that none of the computers on the network could send mail. I even mentioned that none of the 7 users could send mail. At this point, the tech support person is anxious to get me off the phone. After all, the problem is not with the modem, and therefore not with the ISP. Or is it?

By now, you are, I am sure, at the edge of your chair wondering how this issue was resolved.

After much search, gnashing of teeth, and conversations with my favourite tech support person at Premiere System Solutions, the issue was resolved by modifying the MTU setting in the Linksys router the client is using. The number was lowered to 1454.

What is an MTU you ask? BroadbandReports.com provides a brief, but excellent description on their site. Also aptly described on Wikipedia. MTU stands for Maximum Transmission Unit. A conversation on a network between computers is very similar to one between two people. You talk, receive confirmation that the other party has understood what you said when the receiving party nods their head. You then release the next salvo of information. One computer on a network (such as the internet)sends out a piece of data -called a packet- then waits for the other computer to acknowledge receipt of said packet. Once acknowledged, the first computer sends the second packet, and so on, until the conversation is complete and all the data has been sent from the first to the second computer.

A conversation between two people is pretty open and free. One person talks, gets interrupted, and so on. Computers, on the other hand, live in a rigidly regulated world. The computer transmitting the information, gets to do so in packets of a particular size. Routers (which direct traffic traffic from a computer through the internet) have to handle the packet sizes efficiently. Too big, and the router hiccups. Too small, and the router is deluged with traffic. And one compuer may not interrup another. Very civilised.

The optimal size that has been decided upon for the internet is 1500. The routers available at the retail store all default to a packet size of 1492.

So why did we have to change our packet size to 1454?
Why 1454? and not another number?

The second answer first: 1454 was selected because of a conversaton going on in the Speed Guide forum.

Why did we have to change it? My guess is someone changed something at the ISP. Running these networks is a complicated affair. There are many vendors involved all updating diferent parts of the system. Who knows what changes were made where. You can be sure that the frontline workers in tech support were not informed about the changes, not that anyone is ever fully aware of the effect the changes will have.

From start to finish, this process took 3 telephone calls, and 2.5 hours to complete.

Gremlins: 1 September 13, 2007

The computer starts losing time. Windows Media player refuses to play anything. Does not seem to recognize its own presence. Weird and wonderful messages appear on the screen.

The client suspects that the CMOS battery is dead. The computer, as he put it, was going wonky. The clock could not synchronize with any of the many atomic clocks that are out there.

Through the usual set of tests and retests, the problem was eventually narrowed down to the Windows Firewall. It was not turned on, and worse, refused to turn on. No clear message, just would not turn on.

The Microsoft Knowledge Base revealed many articles, but only one that was of particular interest: KB-914230 titled “You cannot turn on or turn off the Windows Firewall setting on a Windows XP Service Pack 2-based computer”. Who comes up with those titles anyways?

The resolution is to repair the Winsock registry key. And no, you don’t have to go into the registry, just repair it through a friendly command prompt: netsh winsock reset. The cause of the problem is that the netfw.inf file is not updated on the computer. The obvious question is: Why not? The answer can be found by searching the Microsoft Technet site. Search for netfw.inf. Have fun.

The computer has to be rebooted once the command is run.

The only problem that remains, is it still loses time, but at least it synchronizes. We now have to establish if the problem is Windows based, or if the crystal that maintains time in the computer has indeed, gone “wonky”.

Update September 18, 2007
You know when you go to a party and ask the I.T. person all sorts of questions? You know when the I.T. person says keep your computer up to date? You know when you ignore all those prompts from Microsoft about updates being ready for your computer? Well, sooner or later, something in your computer will go “wonky”.

In this case, Windows was up to date, but the BIOS was not. Updating the BIOS fixed the issue. Good thing, since I could not find the CMOS battery anywhere! Also took the opportunity to update all the other applications in the computer.

Lenovo has taken a page from Microsoft. They provide a small utility that you can run that will update your computer. Cool.