Is the problem Dixiesys, or is it Comcast?

[Carbo]

I routinely lose access to my email. Right now, for example, I am unable to access or send email. I am running XP Pro, and Outlook Express 6.
I can access my account username@attbi.com without problem. It is the email accounts that I use with my domain that are hanging up every day. In other words, support@mydomain.com, sales@mydomain.com, etc are the problems.
When I try and go directly through SquirrelMail, I am unable to access there, as well.
I have forwarded the error messages and the tracert to both support teams. Apparently, this information is being interpreted differently by each team, because they both point the finger at the other. In the meantime, the customer is stuck in the middle. Any suggestions?

 

[Carbo]

To further clarify, here are the usual error messages I receive:

A time-out occurred while communicating with the server. Account: 'mail.naked-investor.com (1)', Server: 'mail.naked-investor.com', Protocol: POP3, Port: 110, Secure(SSL): No, Error Number: 0x800CCC19

Your server has unexpectedly terminated the connection. Possible causes for this include server problems, network problems, or a long period of inactivity. Account: 'mail.naked-investor.com (2)', Server: 'mail.naked-investor.com', Protocol: POP3, Port: 110, Secure(SSL): No, Error Number: 0x800CCC0F

Your server has unexpectedly terminated the connection. Possible causes for this include server problems, network problems, or a long period of inactivity. Account: 'mail.naked-investor.com (3)', Server: 'mail.naked-investor.com', Protocol: POP3, Port: 110, Secure(SSL): No, Error Number: 0x800CCC0F

Here is the tracert:

1 1 ms 1 ms 1 ms 192.168.0.1
2 13 ms 12 ms 11 ms 10.246.16.1
3 12 ms 11 ms 14 ms bar01-f5-1.plnehe1.fl.attbb.net [66.176.1.33]
4 14 ms 18 ms 10 ms btr01-p2-1.pmbhhe1.fl.attbb.net [66.176.1.89]
5 13 ms 11 ms 12 ms bic01-d6-0.pmbhhe1.fl.attbb.net [66.176.1.121]
6 14 ms 12 ms 15 ms 12.119.94.17
7 17 ms 20 ms 17 ms gbr4-p100.ormfl.ip.att.net [12.123.218.66]
8 * * * Request timed out.
9 27 ms 25 ms 27 ms ggr1-p340.attga.ip.att.net [12.122.12.30]
10 28 ms 28 ms 24 ms att-gw.atl.above.net [192.205.32.182]
11 33 ms 28 ms 28 ms so-6-1-0.cr1.atl2.us.mfnx.net [208.185.0.221]
12 41 ms 38 ms 37 ms 64.124.17.5.tranxact.com [64.124.17.5]
13 46 ms 41 ms 39 ms DV2-L3-ATL-2.dv2.net [209.51.130.211]
14 39 ms 39 ms 40 ms ns5a.dixiesys.com [209.51.150.38]

Trace complete.

Here are the respective replies from dixiesys:

Look at this section of your tracert, it's the path your computer takes to get to Dixie:

6 14 ms 12 ms 15 ms 12.119.94.17
7 17 ms 20 ms 17 ms gbr4-p100.ormfl.ip.att.net [12.123.218.66]
8 * * * Request timed out.
9 27 ms 25 ms 27 ms ggr1-p340.attga.ip.att.net [12.122.12.30]
10 28 ms 28 ms 24 ms att-gw.atl.above.net [192.205.32.182]
11 33 ms 28 ms 28 ms so-6-1-0.cr1.atl2.us.mfnx.net [208.185.0.221]
12 41 ms 38 ms 37 ms 64.124.17.5.tranxact.com [64.124.17.5]
13 46 ms 41 ms 39 ms DV2-L3-ATL-2.dv2.net [209.51.130.211]
14 39 ms 39 ms 40 ms ns5a.dixiesys.com [209.51.150.38]

See hop #8 - it timed out, but, got to hop #13 and #14 (dixie) fine. Where-ever hop 8 is, is where the problem is, and because hop 7 and 9 are both on the ATT.NET network, it's logical to assume that hop 8 is also on ATT. The tracert shows you DID make it to dixie (and therefore Naked-Investor.com) - and obviously you'd get your ATTI mail (it's probably served well before hop #8 ... but wherever that is, is where it's having problems.

and from Comcast:

Thank you for writing to Comcast.

I apologize for any inconvenience this issue has caused you. We only support the connection issues to our servers , which are mail.attbi.com. Your Error message is with your email server mail.naked-investor.com.

To get assistance with the trouble you are having please contact the Customer Support for mail.naked-investor.com.

So, here I am, stuck in the middle. Anyone????

 

[Lord Evermore]

Ignore the hop that's timing out, it's not really timing out. Sometimes a network uses "private" IP space such as 10.x.x.x for internal interfaces, which don't need to be seen directly by anything outside their network. This works fine for inside their network, since the interface only needs to forward packets between itself and other things on that company's network. However it results in nothing outside the network being able to "see" that interface, so trying to traceroute results in a timeout on the traceroute for that hop (nobody's network will forward the packets back to you because they come from a private IP), and once it moves past that hop the router's continue to reply.

If ALL traffic really timed out on that hop, you wouldn't finish the traceroute, and you also would never be able to make a connection.

After the disconnect, are you able to immediately reconnect to the server? Does a traceroute reach the server even when you're unable to connect?

The mail server does respond to pings, so I would advise setting up a command prompt to continuously ping the server (at wide intervals so it's not just constantly pinging) and when the mail client gets disconnected, see whether the pings have stopped. If they have not, that means the server is still reachable, and means that the disconnections are due to the server and not the path to it. The fact that SquirrelMail is unable to reach the mail server as well would seem to indicate that it's a server issue, unless the path between squirrelmail and dixiesys is very much the same as the one to AT&T.

DixieSys doesn't seem to be a very well respected service, bandwidth and accessibility seem to be low. I wouldn't be surprised if this was something on their side. I'm not surprised that they'd pass the problem onto AT&T's shoulder's and seem to have no knowledge of actual routing either, but that's something a LOT of tech support for companies have a problem with these days.

 

[Carbo]

Thanks for the info, Lord. It's a bit high-tech for someone like me, though. I'm going to wait a day or two and see if anything is resolved. I don't know, but I have to think with the error messages pointing to the mail servers on my domain, that this is probably a Dixie issue.
Comcast has further replied twice today:

Upon reviewing the trace route, we do not see a problem. While you do note that hop 8 times out, it could simply be that ICMP is turned off at that hop. After hop 8, there is no discernable problem.

This was followed by:

ICMP is short for Internet Control Message Protocol, an extension to the Internet Protocol (IP) defined by RFC 792. ICMP supports packets containing error, control, and informational messages. The PING command, for example, uses ICMP to test an Internet connection.
By the way, the timeout noted by Cal could be due to the low priority given to ICMP requests by the router. Routers need to take care of data packet routing first, then if enough resources are available it can process ICMP requests.

Does that help, or make any sense?
Maybe it's time to hook up with another host company. Do you recommend any?

 

[Lord Evermore]

The explanation of ICMP is accurate and is just another possible reason for the hop 8 timeout. It only applies when using Windows' tracert command though, because every other traceroute tool uses UDP but Windows uses ICMP. So ICMP response being disabled would make a Windows machine not receive responses there, but other systems would see it. However if the router uses an RFC1918 address (one of the "private" blocks of IPs) then there would be no response no matter what.

The router prioritizing the traffic and responding to ICMP last is a separate thing, configurable by the administrators to make ICMP response the last priority, in order to ensure "real" packets get through first. That's not related to ICMP response being disabled entirely.

I have no experience with any hosting companies so I can't recommdend any. There've been a few threads in these forums I think about that, and certainly there are likely to be many other forums with that sort of thing.