how to test for a qmail virtualdomain?
Does anyone know of a reliable way to programatically test whether an incoming message is part of a qmail virtualdomain or not? I've been looking at qmail-send's environment variables under both a regular setup and a virtualdomain, and can't find a surefire way to differentiate the two cases. Take a look at qmail-command(8) under ENVIRONMENT VARIABLES to see what I'm talking about. I need a method that works under vmailmgr, vpopmail, and qmail's builtin virtualdomains support. For example, under vpopmail, the 'USER' environment variable is set to the virtualdomain instead of the username, so this is easily tested for (if username != $USER, ...). However, this isn't the case under vmailmgr or virtualdomains. Thanks. -- (TMDA - http://tmda.sourceforge.net/) (A qmail-based SPAM reduction system)
Re: how to test for a qmail virtualdomain?
I would say the most reliable way would be to just parse /var/qmail/virtualdomains and compare the LHS with $HOST (although this is a problem while somebody has updated virtualdomains but haven't sent -HUP to qmail-send yet). Do you just need to test for these three packages or does it need to handle all types (for example, maildrop has its own virtual users mechanism)? How reliable does it need to be? Tim On Wed, May 16, 2001 at 09:53:13AM -0600, Jason R. Mastaler wrote: Does anyone know of a reliable way to programatically test whether an incoming message is part of a qmail virtualdomain or not? I've been looking at qmail-send's environment variables under both a regular setup and a virtualdomain, and can't find a surefire way to differentiate the two cases. Take a look at qmail-command(8) under ENVIRONMENT VARIABLES to see what I'm talking about. I need a method that works under vmailmgr, vpopmail, and qmail's builtin virtualdomains support. For example, under vpopmail, the 'USER' environment variable is set to the virtualdomain instead of the username, so this is easily tested for (if username != $USER, ...). However, this isn't the case under vmailmgr or virtualdomains. Thanks. -- (TMDA - http://tmda.sourceforge.net/) (A qmail-based SPAM reduction system)
Re: how to test for a qmail virtualdomain?
Jason R. Mastaler [EMAIL PROTECTED] wrote: Does anyone know of a reliable way to programatically test whether an incoming message is part of a qmail virtualdomain or not? I think you'll have to parse control/virtualdomains. I've been looking at qmail-send's environment variables under both a regular setup and a virtualdomain, and can't find a surefire way to differentiate the two cases. With a virtualdomains entry like: virtual.domain:de5-virtual A message sent to [EMAIL PROTECTED] is very similar to a message sent to de5-foo@localhost. Diff'ing the qmail-command environments, I get: HOST2=sws5.ctd.ornl HOST3=sws5.ctd DTLINE=Delivered-To: [EMAIL PROTECTED] HOST4=sws5 HOST=sws5.ctd.ornl.gov [EMAIL PROTECTED] --- HOST2=virtual HOST3=virtual DTLINE=Delivered-To: [EMAIL PROTECTED] HOST4=virtual HOST=virtual.domain [EMAIL PROTECTED] -Dave
Re: how to test for a qmail virtualdomain?
Tim [EMAIL PROTECTED] writes: I would say the most reliable way would be to just parse /var/qmail/virtualdomains and compare the LHS with $HOST (although this is a problem while somebody has updated virtualdomains but haven't sent -HUP to qmail-send yet). In this case, how about comparing $HOST with control/locals? Someone suggested this in a private message since when qmail rewrites a virtualdomain address into a local one, it changes $USER, but not $HOST. If $HOST is in locals, then it's not a virtualdomain address. Any exceptions to this? Do you just need to test for these three packages or does it need to handle all types (for example, maildrop has its own virtual users mechanism)? All types, I just wasn't aware of maildrop's virtual users mechanism. How reliable does it need to be? Wholly. If not, I'll just make the user set a configuration variable to tell the application to expect a virtualdomain. -- (TMDA - http://tmda.sourceforge.net/) (A qmail-based SPAM reduction system)
