Linux Manpages Online - man.cx manual pages

NAME

/etc/smail/retry − smail retry configuration

THE RETRY FILE

This section details the format and use of the retry control file.

The retry file defines, by target domain, the minimum interval between delivery attempts and/or the maximum duration over which to continue attempting delivery.

Each retry entry should have the form:

    destination[:destination...]  [interval]/[duration]

stating that, for the given destinations(s), delivery will be attempted no more frequently than interval, and that delivery attempts will cease after duration.

The destination is the name that is used for the host retry file which serves both the purpose of preventing Smail from opening more than one simultaneous operation for a given destination and which; and the purpose of recording the reason why a delivery to a given destination failed and when that failure occurred. For all transports except those using the tcpsmtp driver the destination can be either a simple hostname, a fully qualified domain name, or just the proper suffix of a domain name. When Smail is compiled with HAVE_BIND the destinations for transports using the tcpsmtp driver must be specified as an explicit IP address in the format identical to what is generated by inet_ntoa(3) since the host retry lock filenames are the ASCII representations of the IP address being used for the target SMTP server host. This is done so that target hostnames resolving to multiple addresses won’t block all addresses should deliveries to just one of those addresses fail.

When given as a domain name a leading dot (’.’) is ignored but you can use it to document when you really mean a domain suffix in a way similar to the pathalias router driver.

Values for interval and duration are in seconds if no units suffix is given. However, they can be written as a sequence of numbers with suffixes of ’s’ for seconds, ’m’ for minutes, ’h’ for hours, ’d’ for days, ’w’ for weeks and ’y’ for years. For example, ’’10m/2d’’ specifies an interval of ten minutes and a duration of two days; and ’’1h10m30s/1w3d12h’’ would represent an interval of one hour, ten minutes and 30 seconds and a duration of exactly one and one half weeks.

Invalid specifications (attempts to set a negative number, use of invalid units suffixes, etc.) will be logged as configuration errors. Take care not to use a setting that can not be represented internally as a number of seconds greater than the maximum value of an unsigned C integer.

Explicitly setting the interval to zero (’’0’’) will cause it to always be treated as twice whatever is given as the queue run interval with the ’’-q’’ parameter. This will prevent messages to one or more domains with many dead/broken MX targets from blocking the queue with every queue run attempt and will allow other messages in the queue to be processed more quickly. In extreme circumstances this may help avoid accumulating too many queue run processes that are doing nothing useful.

If the duration is set to zero then any message in the queue will only be retried once more before being bounced (i.e. a delivery attempt is made for every message in the queue before it is bounced even if the duration has long since expired or is set to zero).

Either or both of interval and duration may be omitted, in which case they are taken to be the default value specified with the respective global configuration parameter, either retry_interval or retry_duration. Note that the ’’/’’ separator must be present.

As a special case, if a retry file entry has a domain list equal to the special string ’’*’’, that entry matches any domain. This feature is a catch-all which should only be used in the last entry in a retry file. If a retry file contains no ’’*’’ catch-all entry, then the configuration parameters retry_interval and retry_duration will be used to control retries to domains without specific retry entries.

Manpages

NAME

THE RETRY FILE

FILES

SEE ALSO

BUGS

COPYRIGHT