The connection was broken from the receiver’s side. The manual names two causes, and the second is far more telling.
The first — the server was shut down or crashed. An ordinary fault that passes by itself.
The second — the receiver closed the connection in response to the message content. This is an extreme measure: instead of refusing in words, it simply leaves. Filters behave this way when they have nothing to explain or no reason to.
Selectiveness helps tell them apart. Drops on all messages in a row are about the server. Drops on some messages while the rest are delivered normally are about the content.
⚠️ And this is exactly the case where the category misleads: a drop caused by content looks like a network fault but is fixed by editing the message.
We do have an indirect sign: if drops cluster around one mailing while the others go normally, the problem is in its content.
How to tell them apart:
| Observation | Cause |
|---|---|
| all connections to the receiver drop | the server is off or has crashed |
| a part tied to one mailing drops | the message content |
| drops began after a template change | almost certainly the content |
Look for content problems not in the network but in the message. Links, attachments, domains in the text — what a filter looks at.
Talking to the receiver will not help: it said nothing, and there is nothing to explain to it.
# the sign is selectivity, not the disconnect itself pmta show queues --errors example.com
1 relation
3 articles
◇What the node tells about itself and through which channelsSeven observation surfaces: the web monitor, the console in four formats, the log, accounting, accounting search, metrics and export to external analytics.◇The watchdog that restarts the engineKeeps an eye on the engine and brings it back up after a crash by itself, sending a crash report to the postmaster address.◇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.