The number of simultaneous connections: now, the maximum and the allowed limit, in both directions.
What matters is not each value on its own but the distance between the current value and the ceiling. A node that hits the ceiling behaves differently from a node that does not reach it — and is treated differently.
🔴 The outgoing connection ceiling is cut by the license: on the standard one it is always two hundred, no matter what is written in the configuration.
🔴 The ceiling on outgoing connections is cut by the license: on the standard one it is always two hundred, whatever the configuration says.
What to compare with:
| Value | Compared with |
|---|---|
| current value | the usual one for this node, not the neighboring one |
| peak | the ceiling, if there is one |
| growth | the growth rate, not the absolute number |
Fleet nodes are not directly comparable: they have different hardware, different addresses and a different mix of receivers. It makes sense to compare a node with itself over time.
# the metric comes in the status snapshot curl -s "http://127.0.0.1:8080/status?format=json"pmta show status
3 relations
total-max-smtp-outGlobalOne of the configured limits reflected in this number. § 5.3The activation key is a prerequisite for installation, not a step after itThe license key also takes part in the ceiling — so the number is limited not only by the configuration. § 5.3
Throughput peaksConnections have their own maximum mark — together they show how close the node is to the ceiling. § 5.3The whole node state in one documentOne of the numbers in the response: how many connections are open at this moment. § 5.3The receiver went silent mid-conversationEach stalled attempt holds a place in this number for ten minutes. inferenceChange a sending address when messages are already queuedThe overlap lasts exactly as long as the connections opened before the reload live. inference
pmta show status pmta show queues --sort=rcpt --maxitems=20
3 articles
◇What makes a virtual machine riskyRunning on a virtual machine is possible, but the manual does not recommend it without sufficient hardware and properly tuned hypervisor settings: under load the node may stop responding.◇A queue is a receiving server together with a sending addressThe engine groups messages by the “recipient domain plus sending address” pair: one domain from two addresses is two independent queues with their own counters, connections and mode.◇Recipients deferred after a temporary refusalA recipient to whom the receiver responded with 4xx moves into a separate waiting area with a per-receiver interval; two commands can release it early.