Comments on: Some email stuck in Exchange 2003 queue

By: timmarks

timmarks — Thu, 02 Jun 2011 17:06:24 +0000

The four clients all use Outlook 2007, but I think that we might have a misunderstanding. When the clients take the same message that is not going out, and forward it, the forwarded message does not go either. If they just try to resend the same message, it will not go.

Only if they start a new email message, even if they copy the content from the old message and paste it into the new email, it will go. I have not tried to copy the old recipient list and paste it, because I just told them to retype that part. Since I have discovered that starting a new message bypasses the issue, we have no more messages being stuck. I am trying now to see if I can determine the cause.

The failures do not seem to follow any time pattern, just that when the failure occurs, it goes through retries for the next 3 days, never has success, and then sends out an NDR with the following for each recipient in the domain, a separate NDR for each domain:

Your message did not reach some or all of the intended recipients.

Subject: RE: Status
Sent: 05/15/2011 5:54 PM

The following recipient(s) cannot be reached:

a@B.COM on 10/17/2010 5:59 PM
Could not deliver the message in the time limit specified. Please retry or contact your administrator.

b@B.COM on 10/17/2010 5:59 PM
Could not deliver the message in the time limit specified. Please retry or contact your administrator.

Other messages to the same domain are going through fine, no slow downs, it will fail about 3 times with the failing message, then connect to the same server again and send all the other mail in the queue for that domain. The message tracker of System Manager is also of little help, saying “SMTP:Started Outbound Transfer of Message” followed by “Message transferred to *************** through SMTP” once for each failed retry. (Actual *s, not server name or anything that I am removing)

Our main Communication partner also uses Exchange 2003.

Any help is greatly appreciated. I basically had this role thrust upon me a couple of months ago, and have no prior knowledge of Exchange or AD administration other than create a user…

By: guardian

guardian — Thu, 02 Jun 2011 14:12:25 +0000

what email application are you using for those 4 clients are there any other applications clogging the client system when they send the email/s? I’m ruling out restrictions here as you have highlighted that when resent the email is relayed & delivered.

For those messages queued what does the NDR say? Can we rule out AD related problems and DNS here. This could relate to a bigger problem (if it escalates) or could not.
Do these failed deliveries occur randomly or at particular intervals (or time of sending – like sent during the morning) since resending the message is at a different time (after failed delivery)

Monitor the routing engine and remote delivery queue are other message delivery slowing down or are there large sized emails at the same time? Also messages are categorized by exchange for the particular server so this might be cause delay or failure to send the message

What type of mail server does the recipient use?

By: timmarks

timmarks — Thu, 02 Jun 2011 11:54:39 +0000

>Is this happening to all users or just a select few? When mail is sent to other >domains a, b & c are these the same domains the users normally send mail to and >receives from but sometimes messages delivery fails?

So far I have noticed the issue on 4 users in our organization of about 200. Other users also email the same recipients, and even these 4 users are able to communicate with the recipients in some cases. When I had the users start a new email, instead of “forward” or “reply all” on the stuck messages, the new emails went through.

>As mentioned earlier have you communicated this to the remote server admin and >your ISP?

I have contacted the main recipients’ administrator, but we haven’t been able to resolve the issue. I didn’t contact the ISP, mainly because one of the main recipient domains is one with which we have a direct fiber connection with no ISP equipment between.

By: guardian

guardian — Thu, 02 Jun 2011 09:08:54 +0000

Is this happening to all users or just a select few? When mail is sent to other domains a, b & c are these the same domains the users normally send mail to and receives from but sometimes messages delivery fails?

As mentioned earlier have you communicated this to the remote server admin and your ISP?

By: petkoa

petkoa — Wed, 25 May 2011 17:09:30 +0000

> TCP handshake -> SMTP session start -> sender, recipient sent (of previous
> message) -> message data starts -> (series of TCP ACK from other
> server) -> SMTP Reset (ready for next email) -> sender, recipient sent
> (of failing message) -> message data starts -> (TCP ACK + RST from other
> server).

Well, this still could be consistent with graylisting issue – in case if some rate-limiting mechanism is implemented there…

> When the email is sent to multiple domains, it fails on all the domains,
> not just one or two.

This is another thing – odds are against all addressees having “weird” servers… What’s happening if you address message to a list of addresses, one of them erroneous (may be intentionally, for the test)?

By: timmarks

timmarks — Wed, 25 May 2011 12:26:32 +0000

Another thing I notice: When the email is sent to multiple domains, it fails on all the domains, not just one or two. Probably should have mentioned before, but if the recipient list is “user1@a.com; user2@b.com; user3@c.com“, it will fail at a, b, and c, the same way each time, RST and ACK together.

By: timmarks

timmarks — Tue, 24 May 2011 18:18:57 +0000

I understand what you mean, but could this be the case even when there are more than one message in the queue and the previous email message to the same server gets through? I mean:
TCP handshake -> SMTP session start -> sender, recipient sent (of previous message) -> message data starts -> (series of TCP ACK from other server) -> SMTP Reset (ready for next email) -> sender, recipient sent (of failing message) -> message data starts -> (TCP ACK + RST from other server).

By: petkoa

petkoa — Tue, 24 May 2011 18:07:53 +0000

the worst , not the worth, of course ))

By: petkoa

petkoa — Tue, 24 May 2011 18:04:26 +0000

Graylisting is not about blocking senders – there is a “recent connections table” with the connection source IP for the last, say, 2hr (table entry TTL); every incoming connection is looked up in this table and if the source IP is there, mail is accepted – else, a “service temporarily unavailable” is returned and sending server IP is written to the table… If a second connection is made during the TTL of the table entries, the message is accepted. Thus spam bots which are not queuing temporarily returned mail are filtered out.

So if the the queue processing interval is too long (longer than tipical TTL of the entries in graylisting tables) the messages are not going to be delivered – the same, if your server is not properly processing “service temporarily unavailable” message and immediately gives up and returns the message to sender.

For me, your symptoms point to a graylisting issue.

And, BTW, I assume that graylisting – and for that matter, all spam filtering on the servers – is the worth thing that could happen to a internet mail system (not intended to start a flame war, I had a couple of real wars on this subjects with clients).

By: timmarks

timmarks — Tue, 24 May 2011 12:18:10 +0000

Don’t really think it is a graylisting issue, but I could be wrong. I thought that the facts pointed in a different direction. The sender is not blocked, as he can send to the same recipients with the same subject line, as long as it’s a new mail.