Thanks.

Anyone know of a quick daemon that will do this ?

I sort of agree with that, but I feel I should point out that it's not SpamHalter that's writing the console message, it's Mercury core. And Mercury core is only aware that a job in the queue was deleted by a daemon, it doesn't know the reason. 

/Rolf 

Certainly.

I'm submitting 5000 messages in rapid succession using a Java application which uses the javax.mail API.

The messages are small.... less than 1kb.

 I can send and receive low volumes no problem.

Not sure what the temp directory is.... where do I find this in the config? I'm running with the install defaults btw and running the latest  version 4.62.

 Anti-virus.... good thought! Yes, I do have real-time scanning going on. I've excluded the mercury directory and will test again tomorrow.

 Thanks

Under the Pop3 server config, make sure "Offer only unread mail to connected clients" is unticked.

This is really the only Mercury setting that effects this process, it is up to the client which messages to request.

"Mark retrieved mail as read" may or may not have an effect, you will have to experiment.

Depends on your smarthost. Some allow other domain names, some require them to be set up and 'allowed' first (gmail), some will refuse altogether.

MY ISP's mail server has gone off line, if I'm using port 110 it just reports a network error and carries on. If as I was doing, using port 995 and direct ssl it crashes mercury.

 It happens in 4.61 and 4.62, mercury crashes & hogs all the resources.

Error dump available if you need it.

Thanks

Mailing list activity is not shown clearly in the console logs so this could very well be normal. If your list is internal you will not see anything there. If you have MercuryE running and your lists are VERP enabled, you should see activity there for each message flowing out of your system.

As Rolf pointed out, you should upgrade from 4.5x to 4.62 as there are many fixes between the versions.

No.  The problem here is Mercury's forwarding mechanism.  For some obscure reason, it uses the resend strategy to forward mail from the forward file's author to the destination.  For precisely the reasons observed by the problem poster, this is the *wrong thing to do*.  Forwarding should preserve the return path of the sender and introduce an identical message (no modifications other than trace fields) into the queue, changing the recipient to the destination address.  The theory is that, while it is certainly the forward file's author who can do something about any errors, he is as likely to be unable to receive any pertinent error message forwarded to him as to receive the problem message.  For all practical purposes, he does not fall into the delivery path: the person for whom status notices are intended *is* the cause of the status notices, unlike a mailing list, where it makes perfect sense to take responsibility at another address mid-delivery.  It is therefore important that the sender be notified.  The local site configuration problem that resulted in the error condition can't be forwarded reliably, so it must be deposited in a physical mailbox or sent to Postmaster where no forwarding should be occurring or, if it is, it should be to a safe destination.  (Errors are sent with null sender, so there are no further loops if this problem report doesn't reach the Postmaster.)

Cheers,

Sabahattin

You need to use the regular expression option to match against a wildcard. So to catch any file attachment, click "Only check the name portion", put * in the textbox and check "This string is a regular expression".

/Rolf 

You need to put:

server: [192.168.1.10]

in the Local Domains section of Core config

Also you dont need the smtp client module for delivery to local users. 

In that case you would normally get a rejection message from the server, but from your description it appears that the connection is blocked at networking level. But it could be that gmx are a bit over-ambitious in their anti-spam measures. Try contacting them about it, it might work.

/Rolf 

what values do you recomend?

Neither of these are material.  A TCP/IP time out of 300 seconds in MercuryC is a good place to start.

It would also pay to implement an Outgoing filter rule to prevent mail from internal users ("user@sob.intern" or just "user" etc as per their mail client setting) from being sent out to the world.

These would get rejected by anyone checking for MAIL FROM validity (most servers) and any bounces or replies would have nowhere to go.

You could set a rule to delete and/or a reply with template if users need reminding [:)]

You need to upgrade the daemons as well - a lot of changes to the daemon interface occured between 4.5x and 4.6x.

Always make a backup of your system setup before doing an upgrade, so you easily can downgrade if something should arise.

[quote user="Addis"] The problem is that our ISP is rejecting some mailings with 452 #4.5.3 Too many recipients.[/quote]

Start using mailing lists within Mercury. Set to enable VERP processing, and the list submission is split into one message per subscriber. If you enable MercuryB your users can handle the lists via the built in web interface.

Thomas;

Thank you.  While it took a bit to set up this is going to work well for the time being.  Eventually we will turn off acceptance of email for invalid local addresses, but for now this is the answer I was looking for.

Thank you for your help,

Brad Johnson

Hi, I have installed on my Windows XP PC an old Mercury version

(4.01). I want to upgrade to 4.62 version but I have another problem.

One

of Mercury mailbox file is about 2Gb and I noticed that Mercury 4.01

doesn't work correctly. So I tried to deleted mail from Outlook but the

folder doesn't change size. How can I do ?

Install Pegasus Mail on the Mercury/32 server using the same mailbox structure as Mercury/32.   That is if Mercury/32 is use c:\mercury\mail\~n in Configuration | Mercury core "Local mailbox directory path" then Pegasus Mail should be installed to c:\pmail and use c:\mercury\mail when asked for the directory.

You can now access the folders and compress the folder to remove all deleted space.  FWIW, if you install WinPMail Public beta 1 it comes with a utility called mbxmaint.exe that can do this for any mailbox directory you specify.

This (internet name for this system, and local domains) has unfortunately always been rather confusing. Mercury uses the internet name for this system to shape outbound system messages. That fact often creates a "conflict" in response to the answer if MercuryS host (announce myself as) is not entered. The help file is also quite vague here, since the help for MercuryS announce myself as refers to "value in the Mercury My Name field" - that field doesn't exist - but I guess David refers to the Internet name.

What we have done is the following:

Internet name is: mail.praktit.se
MercuryS announce myself is: mail.praktit.se

DNS domain is praktit.se
Host is mail.praktit.se
MX pointer is set to incoming gateway (currently another host)

Domains section therefore read:
mail    praktit.se
mail    mail.praktit.se

The problem is, the bounce messages, that will not return unless the domains section has both entries.

I would have preferred if Internet Name was split into a global setting of Host name, and, server domain be specified separately - but it works the way it is, though ... not clear enough.

The EuroUnion is something else. We have a monetary union (Euro), passport union (Schengen) and the European Community (EC).

The EC doesn't legislate, it is the job of each nation. The EC Council issues directorates, that the countries adopt (to different degrees).

I haven't investigated if the logs are sufficient, but as I recall it they do not. I will meet with Rolf L. to discuss if we can pull the connecting information in a more timely fashion. Since David also said he would re-work the logs I've put this on hold for a more thorough discussion among the testers.

Edit: But in any case, within the EC only ISPs (defined as anyone providing a service for anyone else, in this case an email service for a domain they do not own themselves), have to store this information for x nbr of months.