Autor: John W. Baxter Data: A: Wakko Warner CC: exim-users Assumpte: Re: [exim] An interesting problem (apparent duplicate delivery)
On 9/28/05 9:44 AM, "Wakko Warner" <wakko@???> wrote:
> John W. Baxter wrote:
>> I don't think so, Tom. The topmost Received: header(s) and Delivery-Date:
>> would then differ, and thus the two copies would not have the same MD5
>> checksum, which it is stated they do. I can certainly see your scenario
>> happening elsewhere, though.
>>
>> I'd guess this one comes from a faulty interaction between the MUA and the
>> POP (or IMAP, but POP seems more likely) server.
>
> I have seen this my self and can easily reproduce it (if it's the pop server
> case) If the client is outlook (specifically 2000 as that's what I have
> seen the problem with) and they do not complete the download of the emails,
> the DELETE command is never issued. Outlook likes to delete messages at the
> end of the download. You may not see this if the client is configured to
> leave messages on the server.
Thanks for the reminder...I've encountered it in that case as well.
Furthermore, Outlook Express (at least older versions) left all prior
messages on the server if a message download failed (eg modem dropout
triggered by modem makers who didn't want to license guard time from Hayes).
Fortunately, I haven't had to manually remove such messages for a long time.
>
> What did you use MD5sum against? the user's mail spool on the server (ie
> bypassing the MUA) or the message after it was downloaded?
Not me, but the OP must have seen MD5 outputs from the as-downloaded
messages, since he stated that he couldn't get users to leave the mail on
the server so he could check it.
