random inbound SMTP failures

random inbound SMTP failures

Post by Cal » Wed, 20 Mar 2002 10:22:18



A client has Small Business Server 2000 (integrates Exchange 2000 and ISA).
OS is SP2; Exchange is now SP2; ISA is SP1.

They had complaints of email returning to senders as NDR (various
responses - but most boil down to time to deliver exceeded).

I can open Exchange Admin console and see the sessions on the server; the
connections from the NDR'd ISP's remain open until the 600 seconds are up,
stop, and restart almost immediately.

Mail can flow in and out; though a few connections always seem to fail on
inbound.

The client has a dynamic IP on a DSL connection; DNS is hosted by TZO.com;
the data on the DNS servers always appears to be fine.

Recent changes are the move from one domain name to another domain name
(external).  I cannot for sure say that the problem came after the domain
change, as the use of the service was limited prior to the new name being
set up.

Microsoft has looked at the set-up and says it is OK.

As I said - mail does move out, and most moves in.  Failures seem to be
timeout related.  I can see the sessions time out on the server.  Even short
emails from my system to the server seem to take too long to complete.  No
CPU spikes seem to be occurring.  Trend OfficeScan MailScan service can be
shut down - with no improvement.

Can anyone suggest a path to follow for ultimate resolution to the problem?

Thanks

Cal

 
 
 

random inbound SMTP failures

Post by David Lemson [MS » Tue, 26 Mar 2002 02:22:32


You say they have a dynamic IP, so I can assume that they're using
authenticated TURN from their ISP.  Or are they using POP?  The first thing
I would check is the NDRs: which server is generating them?  If they're
DSNs, then the hostname of the server that generated it will be in <>s in
the human readable part and will be the Reporting-MTA: in the machine
readable part. This will tell you which server can't talk to which server
and is timing out.  It makes some sense to me that there is a chance that
the Exchange server isn't coming in and taking the mail before the
expiration time on the ISP server.

David

--
This posting is provided "AS IS" with no warranties, and confers no rights.

Quote:> A client has Small Business Server 2000 (integrates Exchange 2000 and
ISA).
> OS is SP2; Exchange is now SP2; ISA is SP1.

> They had complaints of email returning to senders as NDR (various
> responses - but most boil down to time to deliver exceeded).

> I can open Exchange Admin console and see the sessions on the server; the
> connections from the NDR'd ISP's remain open until the 600 seconds are up,
> stop, and restart almost immediately.

> Mail can flow in and out; though a few connections always seem to fail on
> inbound.

> The client has a dynamic IP on a DSL connection; DNS is hosted by TZO.com;
> the data on the DNS servers always appears to be fine.

> Recent changes are the move from one domain name to another domain name
> (external).  I cannot for sure say that the problem came after the domain
> change, as the use of the service was limited prior to the new name being
> set up.

> Microsoft has looked at the set-up and says it is OK.

> As I said - mail does move out, and most moves in.  Failures seem to be
> timeout related.  I can see the sessions time out on the server.  Even
short
> emails from my system to the server seem to take too long to complete.  No
> CPU spikes seem to be occurring.  Trend OfficeScan MailScan service can be
> shut down - with no improvement.

> Can anyone suggest a path to follow for ultimate resolution to the
problem?

> Thanks

> Cal


 
 
 

random inbound SMTP failures

Post by Cal » Wed, 27 Mar 2002 00:13:41


Standard SMTP services - both inbound and outbound.  Inbound senders would
send to an SMTP server using the mail.domain.com name


> You say they have a dynamic IP, so I can assume that they're using
> authenticated TURN from their ISP.  Or are they using POP?  The first
thing
> I would check is the NDRs: which server is generating them?  If they're
> DSNs, then the hostname of the server that generated it will be in <>s in
> the human readable part and will be the Reporting-MTA: in the machine
> readable part. This will tell you which server can't talk to which server
> and is timing out.  It makes some sense to me that there is a chance that
> the Exchange server isn't coming in and taking the mail before the
> expiration time on the ISP server.

> David

> --
> This posting is provided "AS IS" with no warranties, and confers no
rights.


> > A client has Small Business Server 2000 (integrates Exchange 2000 and
> ISA).
> > OS is SP2; Exchange is now SP2; ISA is SP1.

> > They had complaints of email returning to senders as NDR (various
> > responses - but most boil down to time to deliver exceeded).

> > I can open Exchange Admin console and see the sessions on the server;
the
> > connections from the NDR'd ISP's remain open until the 600 seconds are
up,
> > stop, and restart almost immediately.

> > Mail can flow in and out; though a few connections always seem to fail
on
> > inbound.

> > The client has a dynamic IP on a DSL connection; DNS is hosted by
TZO.com;
> > the data on the DNS servers always appears to be fine.

> > Recent changes are the move from one domain name to another domain name
> > (external).  I cannot for sure say that the problem came after the
domain
> > change, as the use of the service was limited prior to the new name
being
> > set up.

> > Microsoft has looked at the set-up and says it is OK.

> > As I said - mail does move out, and most moves in.  Failures seem to be
> > timeout related.  I can see the sessions time out on the server.  Even
> short
> > emails from my system to the server seem to take too long to complete.
No
> > CPU spikes seem to be occurring.  Trend OfficeScan MailScan service can
be
> > shut down - with no improvement.

> > Can anyone suggest a path to follow for ultimate resolution to the
> problem?

> > Thanks

> > Cal

 
 
 

random inbound SMTP failures

Post by Cal » Wed, 27 Mar 2002 07:12:02


Sorry

Got interrupted by a power outage...

Standard SMTP services - both inbound and outbound.  Inbound senders send to
an SMTP server using the mail.domain.com name hosted by TZO.COM.  Name
resolution is not the problem as the various delivering SMTP servers can be
seen in the Sessions list.

Reporting-MTA is always the delivering SMTP server.

No POP, no authenticated TURN, no collection from secondary servers.

Thanks!

Cal



> You say they have a dynamic IP, so I can assume that they're using
> authenticated TURN from their ISP.  Or are they using POP?  The first
thing
> I would check is the NDRs: which server is generating them?  If they're
> DSNs, then the hostname of the server that generated it will be in <>s in
> the human readable part and will be the Reporting-MTA: in the machine
> readable part. This will tell you which server can't talk to which server
> and is timing out.  It makes some sense to me that there is a chance that
> the Exchange server isn't coming in and taking the mail before the
> expiration time on the ISP server.

> David

> --
> This posting is provided "AS IS" with no warranties, and confers no
rights.


> > A client has Small Business Server 2000 (integrates Exchange 2000 and
> ISA).
> > OS is SP2; Exchange is now SP2; ISA is SP1.

> > They had complaints of email returning to senders as NDR (various
> > responses - but most boil down to time to deliver exceeded).

> > I can open Exchange Admin console and see the sessions on the server;
the
> > connections from the NDR'd ISP's remain open until the 600 seconds are
up,
> > stop, and restart almost immediately.

> > Mail can flow in and out; though a few connections always seem to fail
on
> > inbound.

> > The client has a dynamic IP on a DSL connection; DNS is hosted by
TZO.com;
> > the data on the DNS servers always appears to be fine.

> > Recent changes are the move from one domain name to another domain name
> > (external).  I cannot for sure say that the problem came after the
domain
> > change, as the use of the service was limited prior to the new name
being
> > set up.

> > Microsoft has looked at the set-up and says it is OK.

> > As I said - mail does move out, and most moves in.  Failures seem to be
> > timeout related.  I can see the sessions time out on the server.  Even
> short
> > emails from my system to the server seem to take too long to complete.
No
> > CPU spikes seem to be occurring.  Trend OfficeScan MailScan service can
be
> > shut down - with no improvement.

> > Can anyone suggest a path to follow for ultimate resolution to the
> problem?

> > Thanks

> > Cal

 
 
 

random inbound SMTP failures

Post by David Lemson [MS » Fri, 29 Mar 2002 06:36:01


So are you saying that TZO.COM tracks their dynamic IP and instantly changes
the DNS record for the company to the current IP address?  And TZO.COM sets
the TTL values on the DNS servers to be zero so that there is no caching
done?  That is the only way I could see that this could work reliably.  But
it seems like maybe you're also seeing problems with hosts that do
successfully look up the name.

I suppose the other thing to check is if there are problems with the inbound
path via DSL, could there be a hardware or other malfunction that makes some
sessions time out?  Is there any pattern for which connecting hosts fail?

--
This posting is provided "AS IS" with no warranties, and confers no rights.

> Sorry

> Got interrupted by a power outage...

> Standard SMTP services - both inbound and outbound.  Inbound senders send
to
> an SMTP server using the mail.domain.com name hosted by TZO.COM.  Name
> resolution is not the problem as the various delivering SMTP servers can
be
> seen in the Sessions list.

> Reporting-MTA is always the delivering SMTP server.

> No POP, no authenticated TURN, no collection from secondary servers.

> Thanks!

> Cal



> > You say they have a dynamic IP, so I can assume that they're using
> > authenticated TURN from their ISP.  Or are they using POP?  The first
> thing
> > I would check is the NDRs: which server is generating them?  If they're
> > DSNs, then the hostname of the server that generated it will be in <>s
in
> > the human readable part and will be the Reporting-MTA: in the machine
> > readable part. This will tell you which server can't talk to which
server
> > and is timing out.  It makes some sense to me that there is a chance
that
> > the Exchange server isn't coming in and taking the mail before the
> > expiration time on the ISP server.

> > David

> > --
> > This posting is provided "AS IS" with no warranties, and confers no
> rights.


> > > A client has Small Business Server 2000 (integrates Exchange 2000 and
> > ISA).
> > > OS is SP2; Exchange is now SP2; ISA is SP1.

> > > They had complaints of email returning to senders as NDR (various
> > > responses - but most boil down to time to deliver exceeded).

> > > I can open Exchange Admin console and see the sessions on the server;
> the
> > > connections from the NDR'd ISP's remain open until the 600 seconds are
> up,
> > > stop, and restart almost immediately.

> > > Mail can flow in and out; though a few connections always seem to fail
> on
> > > inbound.

> > > The client has a dynamic IP on a DSL connection; DNS is hosted by
> TZO.com;
> > > the data on the DNS servers always appears to be fine.

> > > Recent changes are the move from one domain name to another domain
name
> > > (external).  I cannot for sure say that the problem came after the
> domain
> > > change, as the use of the service was limited prior to the new name
> being
> > > set up.

> > > Microsoft has looked at the set-up and says it is OK.

> > > As I said - mail does move out, and most moves in.  Failures seem to
be
> > > timeout related.  I can see the sessions time out on the server.  Even
> > short
> > > emails from my system to the server seem to take too long to complete.
> No
> > > CPU spikes seem to be occurring.  Trend OfficeScan MailScan service
can
> be
> > > shut down - with no improvement.

> > > Can anyone suggest a path to follow for ultimate resolution to the
> > problem?

> > > Thanks

> > > Cal

 
 
 

random inbound SMTP failures

Post by Cal » Sat, 30 Mar 2002 04:19:24


David

It would appear that the issue is being caused by the hardware portion of
the system.  The router in use (D-Link DI-704) was set up in port forwarding
mode.  When it was reset for DMZ mode with the servers IP as the endpoint -
all timeouts disappeared; everything OK.

I had reset the router early on in the diagnostic process, with no
improvement.  Remember that the problem was intermittent in nature - though
there seemed to be a few inbound connecting servers that regularly timed out
on their session connections.

I will now be sure to include a brief test with the DMZ in effect for
communication issues like this.

I also will be testing firmware upgrades for the router to see if that
clears the  issue and I can eliminate the DMZ mode for improved security,
reduced server usage.

Yes - TZO tracks the IP through their server installed client software.
That s/w can check for an IP change as often as desired (default is 10 or 15
minutes) and if different it is posted up to the TZO DNS servers.  The TTL
is not zero on their DNS servers.  I have made a request for information to
them on this issue.  I have not seen this as a non-delivery issue in the
past - though there may be delays in delivery due to this.

Thanks

Cal



Quote:> So are you saying that TZO.COM tracks their dynamic IP and instantly
changes
> the DNS record for the company to the current IP address?  And TZO.COM
sets
> the TTL values on the DNS servers to be zero so that there is no caching
> done?  That is the only way I could see that this could work reliably.
But
> it seems like maybe you're also seeing problems with hosts that do
> successfully look up the name.

> I suppose the other thing to check is if there are problems with the
inbound
> path via DSL, could there be a hardware or other malfunction that makes
some
> sessions time out?  Is there any pattern for which connecting hosts fail?

> --
> This posting is provided "AS IS" with no warranties, and confers no

rights.
 
 
 

1. random inbound email failures, continuous Virtual server sessions

SBS2000 with all current Service Packs (no SQL), most Hotfixes and Security
updates.

Client has rec'd several notices from others that have attempted inbound
email transfers - that have failed for random reasons.  Failed transfers
have involved attachments and mail-lists.  I can send attachments to the
site no problem, and I have yet to see any failed mails to any client at
that site.  I have in fact sent the actual attachment that couldn't get
through, without any failures.

What I can see is that there are always Virtual Server Current Sessions
showing from the same few external IP's (usually 4 to 6 sessions showing).
They hit timeout of 600 secs, go away and restart again,  Some sites have
several connections going at once.

This is NOT a high volume site for either inbound or outbound email.  I have
checked for possible open relay issues - but settings all seem OK and a test
using Sam Spade showed site was ok.

I have turned up logging on the connection; standard logging showed Timeouts
occurring on the IP's associated with the sites showing in the Current
Sessions list.  These are not small sites - Universities and ISPs - so they
should be working properly on their end.

The connection is via DSL, through a router, to an external NIC on the
server.

Any hints or suggestions as to where to look or what to do about it will be
appreciated.

Thanks

Cal

2. HELP! Exporting Data from Active Directory / and Win2000 Server.

3. SMTP Inbound Failure!

4. ETRN request?????

5. inbound SMTP Mail failure

6. Mail enabled contacts

7. SMTP Service = Random emails

8. Urgent, please help

9. Random SMTP shutdown

10. Random SMTP outages

11. Opening and forwarding "inbound mail failure"?

12. Help.....Inbound Notification Failures

13. Two Notification Inbound Mail Failure Messages