I come from Efa (https://efa-project.org/) which is basically using mailscanners as a web interface and all other goodies that proxmox also uses, in the end most products use the same stuff under the hood and come with a different GUI wrapper. I used this for a few years, and it requires a lot of manual configs. But once it works it does its job.
The project is now discontinued, and I wanted something more plug and play I can use as a mail gateway. Proxmox is very easy to install and setup. The GUI makes it very easy.
My impression is that as a general firewall or gateway they both do their jobs but after a few months of using proxmox I noticed its biggest flaw. Efa holds messages which allows it to act as a temporary backup server which is nice. Proxmox does not. Not a big ideal as most email servers will keep retrying but the biggest flaw I see with Proxmox for this reason is that you cannot train messages from the web interface if they passed the filters.
Once configured with lists and general rules, they both stop most hard spam messages but with Mailscanner, after a few weeks of manually training messages that passed the filters, it reduced this to basically 0.
These spam messages cannot be caught with the regular filters because they come from Gmail, outlook, and are changing accounts, mostly low volume, but they clearly have very similar content that can be caught by analyzing the text. The only way to fight them is training the Bayesian filters from spam assassin.
The amount of Spam I receive since I switched to Proxmox is staggering. From 1 a week, to 20 a day. The only reason is that you cannot train this spam messages. I see a lot of people complaining about this for years (since 2018). I see people doing hack around like having an IMAP account to move spam messages, scripts that transfer them to the proxmox server, etc.
Overall I just wanted to say that Proxmox is a nice product, but it's deeply flawed to fight spam. We should not be hacking around scripts to train messages. What is the purpose of all the amazing hard work they made on the UX/GUI if they are then killing one if not the most important of spam assassin which is the training. This makes no sense at all. The Bayesian filters are totally useless if you cannot tell them which messages are spam.
The project is now discontinued, and I wanted something more plug and play I can use as a mail gateway. Proxmox is very easy to install and setup. The GUI makes it very easy.
My impression is that as a general firewall or gateway they both do their jobs but after a few months of using proxmox I noticed its biggest flaw. Efa holds messages which allows it to act as a temporary backup server which is nice. Proxmox does not. Not a big ideal as most email servers will keep retrying but the biggest flaw I see with Proxmox for this reason is that you cannot train messages from the web interface if they passed the filters.
Once configured with lists and general rules, they both stop most hard spam messages but with Mailscanner, after a few weeks of manually training messages that passed the filters, it reduced this to basically 0.
These spam messages cannot be caught with the regular filters because they come from Gmail, outlook, and are changing accounts, mostly low volume, but they clearly have very similar content that can be caught by analyzing the text. The only way to fight them is training the Bayesian filters from spam assassin.
The amount of Spam I receive since I switched to Proxmox is staggering. From 1 a week, to 20 a day. The only reason is that you cannot train this spam messages. I see a lot of people complaining about this for years (since 2018). I see people doing hack around like having an IMAP account to move spam messages, scripts that transfer them to the proxmox server, etc.
Overall I just wanted to say that Proxmox is a nice product, but it's deeply flawed to fight spam. We should not be hacking around scripts to train messages. What is the purpose of all the amazing hard work they made on the UX/GUI if they are then killing one if not the most important of spam assassin which is the training. This makes no sense at all. The Bayesian filters are totally useless if you cannot tell them which messages are spam.