Hello all,

I've developed our spam notification emails, so that recipients can accept & whitelist particular emails in their quarantine by clicking the relavent link without having to log into anything.

Anyhow, some of these email notification are being detected as corrupt, either Outlook can't open some of them, or the backup software (Veritas) reports them as corrupt/bad during the backup run.

I thought the problem was due to line length and thought i'd fixed it, but since my latest update there has since been another corrupt email detected.

To take a look at the RAW email as it is stored in the database visit http://mx1.uksubnet.net/spam2/getmsg.asp?msgid=3392812

This pulls the email from the SF database so what you see is exactly what is sent. You'll see that none of the lines in this email are over 76 characters long, so what else could the problem be?



Edited by lyndonje

Hi Mike,

How much experience do you have with MIME emails? To be honest I have none, this is the first MIME email I've created from code.

I only ask because I got my references from http://mailformat.dan.info/headers/mime.html

Which I think infers this is how the encoding should be - however I understand your logic.

What do you think?

Hi Mike,

I've followed your link, and it looks like my formatting is the same as in example 9.1 (Multipart/alternative inside Multipart/related).

I can't see where it mentions any con's in using 9.1?

Not saying your wrong - because I don't know either way, I'm just a little confused :)

lyndonje wrote: Mike, just to clarify, what were you refering to when you mentioned the boundaries needed to be unique? I know they need to be, but thought mine would be? How do you know if they are/arn't unique?

The "multipart/alternative" and the "multipart/related" parts have to use different boundary strings...

of course, you're right... looks like I need a new pair of glasses...

Here are some more messages that were detected as corrupt in last nights backup:

Message 1
Message 2
Message 3
Message 4
Message 5



Edited by lyndonje

Hi R,

Are you following the links or copying them? They'll only work it the referer passed is www.logsat.com.

If for some reason this can't work for you let me know.

Thanks,
Lyndon.

Got it. I use SSL when browsing the forums, and thus your referrer check blocked me. Looking at the messages, I'll let you know if we spot anything.

Yeah, similar. The problem I see with that is its another step the user has to take to find out if anything genuine has been stopped. They may take the time follow the link only to find it was all spam anyway.

The email I send lists the sender address & subject. If there is nothing of interest in the email the user can simply delete/ignore it. The emails listed in the notification will then automatically be deleted after 7 days, and even if they arn't deleted they won't be mentioned in future notifications. In the tblQuarantine table I've added a notified field, and an auth field. The asp generates a random auth code for each message and sets the notified flag to true when a message has been included in a notification. Therefore the SQL statement in the auto_notify.asp page only pulls messages where deliver, expired and notifed are all false.

The link in the email contains the MsgID, Auth Code and Quarantine ID. When the link for a particular email is followed, an ASP page is run that makes sure the MsgID, Auth Code & Quarantine ID all match. If so its sets the deliver flag to true for SF to whitelist and forward. If all three codes dont match, or an email has already been delivered or deleted, an error is displayed indicating the likely cause. Only problem is some of these are corrupt.... and I don't know why!


Edited by lyndonje

Anybody have anymore ideas on this? Roberto have you managed to take a look?

Just thinking on... incase the email generated and saved in the database isn't corrupt, and the email is being somehow corrupt during transit, does anybody know of a way to see the full and raw source of an email in an exchange mailbox via outlook and compare that with the data saved in the SF database?

Hi Roberto,

Got my hopes up there! But as is often the case, if you're using Exchange this doesn't work, or so the page says:

This works on mail obtained from Internet mail transports, not Exchange server mailboxes.

Just had another idea to give us an indication on whether they are being corrupt in transit... notifications that have already been sent to the respective recipeints that have been detected as corrupt are still stored in the database. I can change the EmailTo field of previously known corrupt emails to myself, and set deliver to 1 and expire to 0. Let them come through to me. If all are again detected as corrupt, its unlikely to be a problem in transit, however if some arn't detected as corrupt, it could be transit?

What do you think?

Another thing I've just tried... The above two mentioned emails which Outlook won't even display, I've set the EmailTo to a POP3 account and downloaded with Outlook Express - both emails downloaded and displayed fine? Would this indicate a problem with these emails and exchange?