Demo
Demo

Knowledge base

PMTA knowledge base›Monitor screen§ 5.5

Receivers summed across all sending addresses

Domains Display
A complement to the queues screen — the same mail, summed differently

The screen shows recipient domains, summing all sending addresses for each. It complements the queue screen: the same mail, but not split into pairs (queue-identity).

Without the filter form this is “Top domains”: the ten receivers with the most recipients — http-default-max-items sets the length.

Four columns, and together they answer the question. Recipients, kilobytes, open connections and the last error: the first three say whether the receiver is being worked with, the fourth how it answers.

The last error column is the most telling. The chapter has a separate section just to explain it: the codes and texts that appear there need decoding.

Filtering by domain is a form on the page itself; handy when there are many receivers and the question is about one.

The main trap is comparing the numbers on this screen with the queues screen. These are different aggregates: here addresses are summed, there they are separated. One domain sent from three addresses gives one line here and three there, and the totals will not match — not because the counters lie, but because the questions are different. The mistake is all the more dangerous because both screens look alike.
What the Hub does. It builds its own per-provider view, and it sums the same way — all addresses together.

⚠️ This is convenient for the question “how are we doing with this receiver overall” and unsuitable for the question “what is happening in a specific queue.” The difference is the same as between the node’s two screens, and they must not be confused: limits apply to a pair, not to a domain, so a decision made from a summed number is applied somewhere other than where it was calculated.

The Hub shows the last error too — but already parsed into a category, not in the receiver’s words.

Recommended values

Which screen for which question:

Question Screen
how are we doing with this receiver overall domains — here
what is happening in a specific queue queues
why the receiver refuses the last error here, then an investigation
which limit to apply queues: a limit applies to a pair

The last line is the most important. A limit decision made from a summed number is applied to each queue separately and is therefore multiplied by the number of addresses.

The last error is the latest one, not the most frequent. One random refusal will push out of the column the cause that is actually getting in the way.

The table matching questions to screens is ours. From the manual here come the set of four columns, selection by domain and the fact that this screen sums addresses while the queues screen does not.
Example configuration
# the same from the command line
pmta show domainspmta show queues --sort=rcpt
The highlighted line sums up the addresses — one line per domain. The line below separates them: a domain sent from three addresses will have three lines there, and the numbers will not match.
yellow — — directive names, blue — blocks. They are highlighted the same way in the text of the card.
▲

Common mistakes and problems

what people get wrong and how it ends3 mistakes▶

Calculating a limit from the summed number

§ 5.5

What it costsThe screen sums addresses, while a limit applies to a pair. A number calculated here and written into a receiver block goes to each queue separately and is multiplied by the number of addresses.
How to spot itThe actual flow to the receiver is several times higher than calculated, and the multiple equals the number of sending addresses.
How to do it rightCalculate the limit from the queues screen. This screen answers a different question.

Reading the last error as the main cause

§ 5.5

What it costsThe column shows the most recent error, not the most frequent one. One random refusal pushes out of it the cause that is actually getting in the way, and the investigation goes astray.
How to spot itThe error text on the screen does not match what is seen in the breakdown of refusals by category.
How to do it rightTake the distribution of refusals from accounting, and read this column as “what happened last.”

Comparing the totals of the two screens

§ 5.5

What it costsThese are different aggregates, and they are not supposed to match. The mismatch is taken for a counter error, and the cause is looked for where there is none.
How to spot itThe total across queues does not add up to the domain line.
How to do it rightRemember which view is in front of you: one sums addresses, the other separates them.
Compatibility
PowerMTA 6.0
described in the manual of this version · platform: any
The vendor does not state the first version for every directive — there is no note here. The name also appears in the 5.0 edition of the manual.
⇄

Related entries

3 relations

Can be replaced by

QueuesThe second view: there the same mail is separated by pairs, not summed. § 5.5Receivers with the largest queueThe same summary, returned by a request. § 5.5

Where you can see it

The receiver went silent mid-conversationOne of the failures that end up in the last error column. § 5.8

Referenced from

A queue is a receiving server together with a sending addressThe second view of the same data: it adds addresses together and gives one line per domain. § 5.5Details of one receiverThe screen that shows the same thing to a human. § 5.5Per-receiver summaryThe same list on the monitor screen. § 6.3.13

How to check on the node
It is useful to see the difference between the two views once:
pmta show domains
pmta show queues --sort=rcpt --maxitems=50
For one domain sent from several addresses, the first command will have one line, the second a line per address. The mismatch of totals is not a counter error but different questions.
◇

What else to read

3 articles

◇How the policy for a receiver is assembledSettings are assembled from the most general pattern to the most precise; the order of blocks in the file means nothing.◇Backoff mode — the queue's second personalityA queue that a receiver has started refusing switches to a parallel set of limits, retries and routes — ten directives that act only in this mode.◇Reputation monitoring inside the nodeThe engine itself watches the receiving servers’ replies and changes the queue mode on a pattern match — a reaction within the connection, not through telemetry.

In the Hub
Cost of a mistake
safe: a wrong value breaks nothing
Scenarios
Troubleshoot deliverability · Tune the pace for a provider