Address verification daemon
qmail-verify
qmail-verify receives UDP packets containing local email addresses and returns a single byte to indicate if the address is valid or invalid to the sender of the UDP packet. qmail-smtpd or qmail-qmtpd are typical clients using the qmail-verify service, although at present only qmail-smtpd has had this functionality added. qmail-verify is based on Paul Jarc's realrcptto patch for qmail (http://code.dogmap.org/qmail/).
qmail-verify uses the files control/locals, control/virtualdomains, users/cdb, the system password file entries (typically in /etc/passwd ) as well as the existence or not of users' home directories and .qmail[-xxx] files to determine if a given address is valid.
Where a qmail system uses .qmail-default files on a per-domain basis in a virtual domains setup, this is likely to result in all addresses being considered 'valid'. This may not in fact be the case in certain situations, such as with extensions/adaptations to qmail like vpopmail which use .qmail-default files throughout (delivery in this case is subsequently handled by a vpopmail component). In these cases a replacement for qmail-verify will be required that can determine address validity.
Other customised qmail installations that use different methods to locate users' mailboxes are likely to need alternatives to qmail-verify or a modified version of it for address verification.
qmail-verify should be invoked as user root to have sufficient privileges to determine the validity of a given address. In certain single-UID virtual domains setups, it may be sufficient to run qmail-verify as the single-UID.
By default, qmail-verify listens on localhost (127.0.0.1) on port 11113. This behaviour can be changed by setting the environment variable LISTEN to specify the IP address and/or port: Set this to the desired IP address, optionally followed by a colon and port, thus for example LISTEN="192.168.1.1:10101".
qmail-verify is implemented by taking the various pieces of qmail that parse an address and combining them in the same executable, qmail-verify. Thus logic is taken from qmail-send, qmail-lspawn, qmail-getpw and qmail-local.
The incoming packet contains just the email address to be checked as a string. The string is optionally terminated with a 0 byte.
The response packet contains a single byte to indicate whether the address is valid. The lowest-order bit of this byte indicates the result: 0 for 'valid', 1 for 'invalid'. Other bits of this response byte are set by qmail-verify to give further debugging information; these other bits should generally be disregarded.
Although not especially designed as a new protocol, extensions to qmail-verify could require the query string to be 0 terminated to separate it from other data to follow. Currently the response packet contains the response byte and the 'Controlling user'; more information could potentially be returned if required.
At startup qmail-verify reads the following qmail control files: control/envnoathost, control/locals, control/percenthack, control/virtualdomains. If changes are made to any of these files, qmail-verify should be restarted for the changes to take effect in qmail-verify.
If you are using different machines for qmail-verify and qmail-smtpd you should ensure that the machine providing the qmail-verify service has a full set of control files as well as the mailboxes; the machine running qmail-smtpd still needs control/rcpthosts to be setup.
qmail-verify logs each decision it makes to stderr: The address followed by whether it's valid or not.
Andrew Richards, building on the work of Paul Jarc and Dan Bernstein, and with plenty of help along the way from Russell Nelson, John Levine and Charles Cazabon amongst others.