Backoff, urgentbackoff, normalbackoff, nonurgentbackoff, ipbackoff Channel Options

From Messaging Server Technical Reference Wiki
Jump to: navigation, search

Delivery retry intervals (backoff, urgentbackoff, normalbackoff, nonurgentbackoff, ipbackoff)

Backoff options specify the frequency of message delivery retries when messages aren't successfully delivered the first time. These options all accept a series of intervals as arguments. The first interval specifies the time to wait before the first retry, the second specifies the time to wait for the second retry, and so on. The last value given specifies the time to wait for all subsequent retries. Up to eight intervals can be specified. Deliveries are attempted for a period of time specified by the notices channel option. Delivery will fail if successful delivery cannot be made within the time allowed by the last notices channel option setting.

Interval values use ISO 8601 periodic time syntax:


years, months, weeks, days, hours, minutes and seconds are all integer values.

Note that all of the letters in the value must be written in upper case.

Separate interval settings are allowed for messages with different priority settings. The urgentbackoff channel option sets the retry intervals for urgent messages, normalbackoff sets the retry intervals for normal priority messages, and nonurgentbackoff sets the interval for non urgent messages. backoff sets the retry intervals for all messages regardless of priority.

Since backoff sets the retry interval for all priorities, it's setting overlaps with the other options. In legacy configurations the order in which the options appear in the channel definition determines the result: The last option specified determines the result. In unified configuration prior to MS only one option affecting messages at a given priority can be specified, so setting backoff will remove all the other options from the channel. In MS and later any mixture of options can be specified, with the more specific options all overriding the more general backoff option for messages at a given priority.

Additionally, as of MS, an additional ipbackoff option is provided to set backoff times for messages that have been placed in IP backoff mode. The ability to set ipbackoff values on a per-domin basis is also provided by the ipbackoff smartsend parameter.

The default intervals between delivery retry attempts in minutes are:

urgent: 30, 60, 60, 120, 120, 120, 240
normal: 60, 120, 120, 240, 240, 240, 480
nonurgent: 120, 240, 240, 480, 480, 480, 960
ip: 60, 120, 120, 240, 240, 240, 480

Note that setting the various options on the defaults channel can be used to override these built-in defaults, as opposed to having to set the options on every channel with an associated queue in the configuration, e.g.,

msconfig> set channel:defaults.backoff "PT30M PT1H PT2H PT4H"

Also note that the MTA has special handling of the case of problems delivering to some recipients of multi-recipient messages: in such cases, the failing recipients are eligible for another "immediate" delivery attempt without regard to *backoff setting; see the discussion of MTA message transaction log file "Z" records.

Note that ims-ms channels and LMTP client TCP/IP channels have special case handling that overrides normal backoff for the specific error condition of encountering IMAP_MAILBOX_LOCKED when attempting delivery to a Message Store user.

See also: