Demo
Demo

Knowledge base

PMTA knowledge base›Block Global›Directive§ 3.3.1.2

total-max-smtp-out

Globalcriticalset by the Hub

The ceiling of outgoing connections for the whole node
in the configuration this is a linetotal-max-smtp-out 1200 (Enterprise licence), 200 (Standard licence)inside the block Global
The ceiling of outgoing connections for the whole node

How many outgoing SMTP connections the node holds at once, in total across all queues and all sending addresses. This is the upper bound under which all other connection limits work.

It kicks in first and silently. If a receiver block has max-smtp-out 200 and the node ceiling is lower, only as many connections open as the ceiling allows. The configuration still says 200, and the setting looks applied. On a Standard license the ceiling is also limited by the license itself.

The real limit is not here. The manual warns separately: the engine does not check the upper bound, and everything runs into the machine’s descriptors and memory. A connection costs a descriptor, and when they run out, the failures come not from the receiving server but from the operating system.

The main trap. The ceiling is shared among all queues, not given to each. A fleet of ten addresses sending to five receivers is fifty queues competing for the same pool of connections. When it runs short, queues do not fail, they wait — and it looks like “slow” rather than “hit the limit.”
What the Hub does. It sets 5000 — on an ordinary fleet this is hard to hit, and the headroom removes a whole class of unexplained slowdowns. The value does not depend on the chosen pace: pace is controlled by the queue limits, and the node ceiling only guards against resource exhaustion. The Hub does not read the license tier, so it has no way to warn that two hundred applies on the standard one.
Recommended values

Our value is 5000. The vendor recommends going no higher than 7500 and reminds you that the real limit is the machine’s descriptors and memory, not this line.

The license matters more than what is written. Standard allows no more than two hundred simultaneous outgoing connections, and everything above that is dropped without a message. Enterprise takes 1200 when the line is absent.

Count from the fleet, not from the node. The ceiling is shared by all sending addresses and all queues: ten addresses at fifteen connections each to ten receivers make fifteen hundred, and such a fleet on a standard license hits two hundred long before it exhausts its own limits.

The defaults, the standard license ceiling and the advice not to exceed 7500 are the vendor’s. The value of 5000 is ours.
Example configuration
# node ceilings — outside any blocks
total-max-smtp-out 5000total-max-smtp-in 1000
That is how these two lines stand in the configuration the Hub writes — side by side, at the very top of the file.
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▶

Raising the queue limit without raising the node ceiling

§ 3.3.1.2

What it costsThe instance limit kicks in first, and on a Standard license this happens without a message — the config says 200, but only as many work as the ceiling allows.
How to spot itThe number of connections hits the same value regardless of what the receiver block says.
How to do it rightRaise the node ceiling first, then the queue limits. Check with the command that shows the effective settings, not by reading the configuration.

Assuming the number in the configuration is the effective ceiling

§ 3.3.1.2

What it costsOn a standard license everything above two hundred is dropped silently: no parsing error, no line in the log. A node with 5000 in the file behaves like a node with 200, and the configuration offers no explanation.
How to spot itThe file has one number, while the node status prints a different, noticeably smaller one in the maximum column.
How to do it rightCheck against the node status output and know the license tier; on a standard one, plan the fleet from two hundred connections, not from the written number.

Raising the ceiling above what the machine can handle

§ 3.3.1.2

What it costsEvery connection costs a descriptor and memory, and the engine does not check them. On a node with a soft limit of a thousand descriptors, a ceiling of five thousand is a promise the machine will not keep.
How to spot itConnections stop opening while the receiving server is alive, and the log shows operating-system-level errors rather than SMTP rejections.
How to do it rightLook at the process descriptor limit on the node and keep the ceiling below it, with headroom for inbound mail, logs and queue files.
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.
Reference data
Syntax
total-max-smtp-out <number>
Values
a positive integer, the number of simultaneous connections; on a Standard license no more than 200 applies; on Enterprise there is no upper limit, the vendor advises going no higher than 7500
Default
1200 (Enterprise licence), 200 (Standard licence) § 3.3.1.2
Applying
reload, no restart needed
Queue
will not touch what is already queued
Where to write it
Global
How many times
once in the block
Everything here comes from the manual, with the section.
Inheritance
There is no inheritance: this is a property of the instance as a whole. Below it are the sending address ceiling and the receiver queue limit, both with the same name (max-smtp-out); above it is the license tier, which is what cuts this line. It is picked up by pmta reload.
⇄

Related directives

4 relations

Depends on

The activation key is a prerequisite for installation, not a step after itThe license tier cuts the effective value down to two hundred on a standard license — silently, with no error or warning. § 3.3.1.2How to turn a sending plan into hardwareThe number of connections spends memory and descriptors, and that is exactly what the sizing estimate counts. inference

Affects

max-smtp-out<domain>The queue limit that works under this node-wide ceiling. § 3.3.1.2max-smtp-out<virtual-mta>The same ceiling, but per sending address: it too works under this node-wide limit. § 3.3.1.2

Referenced from

Four ways to say “slower”The connection ceiling for the whole node: it is also the upper boundary that all connection limits together will hit. § 3.3.1.2queue-priority<domain>Priorities start to work exactly when this ceiling is reached. § 3.3.1.2ConnectionsOne of the configured limits reflected in this number. § 5.3

How to check on the node
The node status prints the effective value, not the written one:
pmta show status
The Connections section, the Outbound row, the maximum column — that is the ceiling actually applied. On a standard license it will read 200 no matter what the configuration says; the neighboring active shows how close you have come to it. § 6.3.17
◇

What else to read

3 articles

◇Different domains in one queue, because the gateway is the sameDomains whose lowest-priority MX records match the list are combined into one queue and limited as a single receiver.◇The activation key is a prerequisite for installation, not a step after itWithout an activation key from SparkPost support the engine neither installs nor runs, and the license tier sets the outbound connection ceiling: on Standard everything above 200 is ignored.◇Cloud restrictionsClouds restrict outbound mail on port 25 until you ask them to lift the limit, and mailbox providers treat their address ranges with suspicion.

In the Hub
Managed at
Configurations → Delivery limits
Value
The Hub picks it itself — pace, warm-up stage, metrics
Cost of a mistake
reversible: harms only while it is set
Scenarios
Capacity planning · Tune the pace for a provider
Priority
critical
Neighboring settings of the block Global:max-memory-usage-percentip-connection-limitmx-connection-limitsmtp-proxy-source-hostdomain-keyAll 108 directives of the block →