Weird email problem over wireless - Jeff, got yer ears on?

This started a couple of days ago and it's happening from several
computers at several locations. Can't download email over wireless
connections, get a server time out error (error message pasted in
below). The mail server, SendMail on RedHat 8, is at my office. If I
check mail from any of several comps at the office over the LAN, no
problems. The wireless connections go through a few links, two ends of a
backhaul, then the AP, then the CPE - all smartBridges equipment.

Other than email, all other activities zoom along at a satisfying pace.
I don't know of any reason an AP would knock out POP3 stuff, has me
puzzled right now. Not even sure it's related to the wireless side of
things but it seems more than coincidental.

Suggestions?

Error message from Outlook Express -
Your server has unexpectedly terminated the connection. Possible causes
for this include server problems, network problems, or a long period of
inactivity. Account: 'mail.mydomain.com', Server: 'mail.mydomain.com',
Protocol: POP3, Port: 110, Secure(SSL): No, Socket Error: 10053, Error
Number: 0x800CCC0F

On Tue, 04 Jan 2005 20:44:00 -0500, Rôgêr <(E-Mail Removed)> wrote:

Why me? This has little to do with wireless.

>This started a couple of days ago and it's happening from several
>computers at several locations.

Well, what was changed last? You do keep server logs?

>Can't download email over wireless
>connections, get a server time out error (error message pasted in
>below). The mail server, SendMail on RedHat 8, is at my office. If I
>check mail from any of several comps at the office over the LAN, no
>problems. The wireless connections go through a few links, two ends of a
>backhaul, then the AP, then the CPE - all smartBridges equipment.

A "few links"? Could you be more specific? Anything inline that
would block ports like a router perhaps?

>Other than email, all other activities zoom along at a satisfying pace.
>I don't know of any reason an AP would knock out POP3 stuff, has me
>puzzled right now. Not even sure it's related to the wireless side of
>things but it seems more than coincidental.

Well, the first step to solving a problem is to blame someone. See if
anyone in the office will volunteer. If not, assign the blame to the
office masochist (every office has one).

>Error message from Outlook Express -

Barf. OE and Ouchlook both trap the original error message, try to
interpret it, and give you a bad guess as to what MS thinks it means.
If the message came from an Microsoft mail server, you have a chance
of figuring it out. If from some other server, it's a crap shoot.

>Your server has unexpectedly terminated the connection.

Ok. That means the POP3 server replied with a message similar to "go
away".

>Possible causes
>for this include server problems, network problems, or a long period of
>inactivity. Account: 'mail.mydomain.com', Server: 'mail.mydomain.com',
>Protocol: POP3, Port: 110, Secure(SSL): No, Socket Error: 10053, Error
>Number: 0x800CCC0F

Digging with Google shows:
http://support.microsoft.com/?kbid=813514
http://www.tiscali.co.uk/help/email/...x800ccc0f.html

So much for the official line. On to better ways.

1. Use telnet to get the original error message. Here's a sample
POP3 session:

telnet mail.cruzio.com 110
+OK POP3 Ready mail.cruzio.com 0001b54a
user jeffl
+OK USER jeffl set, mate
pass xxxxxxxxxxxxxxxx
+OK jeffl has 10 visible messages (0 hidden) in 33288 octets.
+OK 10 visible messages (33288 octets)
stat
+OK 10 33398
list
1 1752
2 1888
3 1753
4 1787
5 1755
6 2430
7 2270
8 2921
9 1459
10 15273
..
quit
+OK Pop server at bmail.cruzio.com signing off.

2. I've had chronic problems with MTU discovery issues. Download
fping 2.09 from
http://www.kwakkelflap.com/downloads.html
and ping your mail server with various size packets. For 1500Byte
MTU, the max size is 1472.

C:\>fping 192.168.1.1 -f -s 1472
Pinging 192.168.1.1 with 1472 bytes of data every 1000 ms:
Reply[1] from 192.168.1.1: bytes=1472 time=0 ms TTL=150
(trimmed...)

C:\>fping 192.168.1.1 -f -s 1473
Pinging 192.168.1.1 with 1473 bytes of data every 1000 ms:
Packet size exceeds MTU and needs to be fragmented but DF set.
(trimmed...)

Note that 1472 works, but 1473 generates an error. If you're getting
much small MTU's, you may have a connectivity issue.

3. Check if the mail server is running some kind of firewall
(IPChains, IPFilters, etc) and see if it's blocking port 110 for some
reason? Same with any intrusion detection system (IDS) or filtering
software.

Meanwhile, you're on your own. My truck has an engine problem and it
appears that I'm gonna be busy either doing a repair or finding a
replacement.


--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 AE6KS 831-336-2558

Jeff Liebermann wrote:

> Why me? This has little to do with wireless.

My thoughts at first too, but research shows otherwise.

> Well, what was changed last? You do keep server logs?

I've changed socks and undies, nothing about the network or server
changed that I'm aware of. Maybe something let the smoke out of some
wireless equipment.

> A "few links"? Could you be more specific? Anything inline that
> would block ports like a router perhaps?

No routers between the server and my computer. But there is the backhaul
to the tower and then the main access point. The backhaul is two APs
configured for that purpose. Has worked just fine for nearly two years.

> Well, the first step to solving a problem is to blame someone. See if
> anyone in the office will volunteer. If not, assign the blame to the
> office masochist (every office has one).

I guess my wife will have to do.

> OE and Ouchlook both trap the original error message, try to
> interpret it, and give you a bad guess as to what MS thinks it means.
> If the message came from an Microsoft mail server, you have a chance
> of figuring it out. If from some other server, it's a crap shoot.

It's a RedHat 8 server running SendMail. My Linux guy (not me) says the
server is working fine. Research says he's right. I associated my home
CPE with another access point that is not wireless backhauled and email
worked just fine. Switched back to the backhauled AP and email falls
down. Everything else buzzes right along.

> 2. I've had chronic problems with MTU discovery issues. Download
> fping 2.09 from
> http://www.kwakkelflap.com/downloads.html
> and ping your mail server with various size packets. For 1500Byte
> MTU, the max size is 1472.
>
> C:\>fping 192.168.1.1 -f -s 1472

Reply[1] from 192.168.1.1: bytes=1472 time=20 ms TTL=64
Reply[2] from 192.168.1.1: bytes=1472 time=20 ms TTL=64
Reply[3] from 192.168.1.1: bytes=1472 time=20 ms TTL=64
Reply[4] from 192.168.1.1: bytes=1472 time=20 ms TTL=64
(IP edited to protect the innocent)

> C:\>fping 192.168.1.1 -f -s 1473
> Pinging 192.168.1.1 with 1473 bytes of data every 1000 ms:

Packet size exceeds MTU and needs to be fragmented but DF set.
Packet size exceeds MTU and needs to be fragmented but DF set.
Packet size exceeds MTU and needs to be fragmented but DF set.
Packet size exceeds MTU and needs to be fragmented but DF set.

> Note that 1472 works, but 1473 generates an error. If you're getting
> much small MTU's, you may have a connectivity issue.
>
> 3. Check if the mail server is running some kind of firewall
> (IPChains, IPFilters, etc) and see if it's blocking port 110 for some
> reason? Same with any intrusion detection system (IDS) or filtering
> software.

Nothing changed on the server from a few days before and as I said, my
Linux guru says it's okay, logs say everybody should be happy. And
unlike me, he knows what he's talking about.

> Meanwhile, you're on your own. My truck has an engine problem and it
> appears that I'm gonna be busy either doing a repair or finding a
> replacement.

Good luck on the truck, and seriously, thanks for the pointers. I guess
it's going to be a matter of swapping out equipment (for me, not you)
one piece at a time. That is, unless the guys on the smartBridges list
have a suggestion.

On Wed, 05 Jan 2005 00:23:54 -0500, Rôgêr <(E-Mail Removed)> wrote:

>> Well, what was changed last? You do keep server logs?

>I've changed socks and undies, nothing about the network or server
>changed that I'm aware of.

Well, something has changed. I suggest you interrogate all the usual
suspects.

>No routers between the server and my computer. But there is the backhaul
>to the tower and then the main access point. The backhaul is two APs
>configured for that purpose. Has worked just fine for nearly two years.

Two access points playing bridge? Can't work. What manner of
hardware? What's a "smartbridge"? (My father warned me to stay away
from products with superlatives such as super, amazing, magic,
miracle, etc. I've added smart, intelligent, advanced, and pro to the
list.) Oh...
http://www.smartbridges.com/products/backhaul.asp
Very nice.

>It's a RedHat 8 server running SendMail. My Linux guy (not me) says the
>server is working fine.

Assumption? How about doing the POP3 test I previous posted. It's
easy, painless, and very useful.

>Research says he's right. I associated my home
>CPE with another access point that is not wireless backhauled and email
>worked just fine. Switched back to the backhauled AP and email falls
>down. Everything else buzzes right along.

OK. It's the backhaul radios. Did some dingbat set the remote
management port number to port 110? Instant block if that's what
happened. Maybe someone setup a MAC address filter and left out the
mail server. See if ping, traceroute, telnet, ftp or some other
non-POP3 program works to the mail server.

>Good luck on the truck, and seriously, thanks for the pointers. I guess
>it's going to be a matter of swapping out equipment (for me, not you)
>one piece at a time. That is, unless the guys on the smartBridges list
>have a suggestion.

Bah. Power cycle the backhaul radios and everything plugged into them
first. However, I agree that it's "something" in the backhaul.
However, I don't have a clue exactly what's busted.



--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 AE6KS 831-336-2558

Jeff Liebermann <(E-Mail Removed)> wrote:
> Meanwhile, you're on your own. My truck has an engine problem and it
> appears that I'm gonna be busy either doing a repair or finding a
> replacement.

I think you need a new truck, judging from the number of repairs you've
made to it since last August.

--
---
Clarence A Dold - Hidden Valley (Lake County) CA USA 38.8-122.5

On Wed, 5 Jan 2005 18:15:58 +0000 (UTC), (E-Mail Removed)
wrote:

>Jeff Liebermann <(E-Mail Removed)> wrote:
>> Meanwhile, you're on your own. My truck has an engine problem and it
>> appears that I'm gonna be busy either doing a repair or finding a
>> replacement.

>I think you need a new truck, judging from the number of repairs you've
>made to it since last August.

Not exactly wireloose. Actually, it's the same problem that's been
driving me nuts for the last 9 months. Today, I finally found the
problem. The coil windings in the two glow plug relay ($70/ea) became
corroded from water incursion and shorted. The solenoids would work,
but draw way too much current. The result was plenty of weird and
non-obvious symptoms. Eventually the stupid solenoid stuck closed,
blew a small relay in the controller box ($8), and fried all 4 glow
plugs ($25/ea), which presented an obvious clue. Parts are on order
and the demise of my truck is temporarily postponed.

--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 AE6KS 831-336-2558

Jeff Liebermann <(E-Mail Removed)> wrote:
> On Wed, 5 Jan 2005 18:15:58 +0000 (UTC), (E-Mail Removed)
> wrote:

>>I think you need a new truck, judging from the number of repairs you've
>>made to it since last August.

> Not exactly wireloose. Actually, it's the same problem that's been
> driving me nuts for the last 9 months. Today, I finally found the
> problem. The coil windings in the two glow plug relay ($70/ea) became
> corroded from water incursion and shorted. The solenoids would work,

Nah. Once they go over the edge, the little problems all merge and look
like one. In August it was the power steering, then it was something that
required duct tape (which I presume was not the power steering. I don't
think duct tape would hold very long in the face of power steering fluid).

I try to avoid buying more than one set of tires for most of my vehicles.
I slipped up a bit on one of them. It's had lots of tires, and all of the
accouterments replaced at least once (power steering, water pump, fan
clutch, clutch, alternator, U-joints, A/C compressor, smog pump).
It has been reduced to garbage hauling and resting quietly in the driveway.


--
---
Clarence A Dold - Hidden Valley (Lake County) CA USA 38.8-122.5

On Thu, 6 Jan 2005 06:47:05 +0000 (UTC), (E-Mail Removed)
wrote:

>Nah. Once they go over the edge, the little problems all merge and look
>like one. In August it was the power steering, then it was something that
>required duct tape (which I presume was not the power steering. I don't
>think duct tape would hold very long in the face of power steering fluid).

JB Weld, layer of sheet rubber, and two hose clamps. It worked long
enough to make it to the repair shop. We ordered a junk gearbox which
had a much smaller leak in exactly the same location. Oops. So, I
magnifluxed the crack, filed it open, and brazed it shut with my
torch. I had to keep it half under a bucket of water to keep the heat
from exploding power steering fluid and melting the gaskets. Working
fine now.

I drove it home and it wouldn't start the next morning. Clogged
diesel fuel filter. Easy fix but took me all day to find it. When
the filter started working correctly, the pressure on the backflow
hose increased blowing out the hose. I was leaving puddles of diesel
in the parking lot before I figured out what was happening. Shoving a
piano wire down the hose toward the gas tank removed the blockage.

>I try to avoid buying more than one set of tires for most of my vehicles.
>I slipped up a bit on one of them. It's had lots of tires, and all of the
>accouterments replaced at least once (power steering, water pump, fan
>clutch, clutch, alternator, U-joints, A/C compressor, smog pump).
>It has been reduced to garbage hauling and resting quietly in the driveway.

I've already replaced the water pump so that's not going to blow. I
have three alternators that I rebuild as needed. I used much bigger
and higher voltage diodes on the last rebuild. It's holding so I
don't expect much trouble there. It's a diesel so no smog pump. No
A/C. Power steering was recently fixed. Turbo is in good shape.
Clutch was replaced when I swapped out the tranny. I did the brakes
last year. Not much left to break. Well, the injectors have 150K
miles on them, so I guess it's time. I have two sets so it's off to
the rebuilders. My only complaint is that I lost a few days of work
and that I hate working in the rain with a cold.

One big advantage to fixing trucks to fixing electronics. With the
truck we can *SEE* what's leaking, cracked, or smoking. With
electronics, we're no better than blind (without test equipment).

Bringing off topic discussions to a new low...


--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 AE6KS 831-336-2558

Jeff Liebermann <(E-Mail Removed)> wrote:
> One big advantage to fixing trucks to fixing electronics. With the
> truck we can *SEE* what's leaking, cracked, or smoking. With
> electronics, we're no better than blind (without test equipment).

I keep the old truck around just as a reminder of what "shade tree
mechanics" could do. My newer cars are unrepairable black boxes outside
the dealer.

--
---
Clarence A Dold - Hidden Valley (Lake County) CA USA 38.8-122.5

Jeff Liebermann wrote:
> On Tue, 04 Jan 2005 20:44:00 -0500, Rôgêr <(E-Mail Removed)> wrote:
>
> Why me? This has little to do with wireless.
>
>
>>This started a couple of days ago and it's happening from several
>>computers at several locations.
>
>
> Well, what was changed last? You do keep server logs?
>
>
>>Can't download email over wireless
>>connections, get a server time out error (error message pasted in
>>below). The mail server, SendMail on RedHat 8, is at my office. If I
>>check mail from any of several comps at the office over the LAN, no
>>problems. The wireless connections go through a few links, two ends of a
>>backhaul, then the AP, then the CPE - all smartBridges equipment.
>
>
> A "few links"? Could you be more specific? Anything inline that
> would block ports like a router perhaps?
>
>
>>Other than email, all other activities zoom along at a satisfying pace.
>>I don't know of any reason an AP would knock out POP3 stuff, has me
>>puzzled right now. Not even sure it's related to the wireless side of
>>things but it seems more than coincidental.
>
>
> Well, the first step to solving a problem is to blame someone. See if
> anyone in the office will volunteer. If not, assign the blame to the
> office masochist (every office has one).
>
>
>>Error message from Outlook Express -
>
>
> Barf. OE and Ouchlook both trap the original error message, try to
> interpret it, and give you a bad guess as to what MS thinks it means.
> If the message came from an Microsoft mail server, you have a chance
> of figuring it out. If from some other server, it's a crap shoot.
>
>
>>Your server has unexpectedly terminated the connection.
>
>
> Ok. That means the POP3 server replied with a message similar to "go
> away".
>
>
>>Possible causes
>>for this include server problems, network problems, or a long period of
>>inactivity. Account: 'mail.mydomain.com', Server: 'mail.mydomain.com',
>>Protocol: POP3, Port: 110, Secure(SSL): No, Socket Error: 10053, Error
>>Number: 0x800CCC0F
>
>
> Digging with Google shows:
> http://support.microsoft.com/?kbid=813514
> http://www.tiscali.co.uk/help/email/...x800ccc0f.html
>
> So much for the official line. On to better ways.
>
> 1. Use telnet to get the original error message. Here's a sample
> POP3 session:
>
> telnet mail.cruzio.com 110
> +OK POP3 Ready mail.cruzio.com 0001b54a
> user jeffl
> +OK USER jeffl set, mate
> pass xxxxxxxxxxxxxxxx
> +OK jeffl has 10 visible messages (0 hidden) in 33288 octets.
> +OK 10 visible messages (33288 octets)
> stat
> +OK 10 33398
> list
> 1 1752
> 2 1888
> 3 1753
> 4 1787
> 5 1755
> 6 2430
> 7 2270
> 8 2921
> 9 1459
> 10 15273
> .
> quit
> +OK Pop server at bmail.cruzio.com signing off.
>
> 2. I've had chronic problems with MTU discovery issues. Download
> fping 2.09 from
> http://www.kwakkelflap.com/downloads.html
> and ping your mail server with various size packets. For 1500Byte
> MTU, the max size is 1472.
>
> C:\>fping 192.168.1.1 -f -s 1472
> Pinging 192.168.1.1 with 1472 bytes of data every 1000 ms:
> Reply[1] from 192.168.1.1: bytes=1472 time=0 ms TTL=150
> (trimmed...)
>
> C:\>fping 192.168.1.1 -f -s 1473
> Pinging 192.168.1.1 with 1473 bytes of data every 1000 ms:
> Packet size exceeds MTU and needs to be fragmented but DF set.
> (trimmed...)
>
> Note that 1472 works, but 1473 generates an error. If you're getting
> much small MTU's, you may have a connectivity issue.
>
> 3. Check if the mail server is running some kind of firewall
> (IPChains, IPFilters, etc) and see if it's blocking port 110 for some
> reason? Same with any intrusion detection system (IDS) or filtering
> software.

Another really useful tool is tcptraceroute if you have a *nix box
around. Works well to indicate breakpoint when Telneting to port fails.

http://michael.toren.net/code/tcptraceroute/


Also Ethereal to see at which point in the conversation the failure is
occurring. such as during auth or sending data.

http://www.ethereal.com/


John