Lack of resources on the ML server

Quack, Sandro noticed some posts did not appear quickly on the interface. There's quite a lot of traffic and it was not keeping up, so I added a vCPU (2->3), but I had to restart the VM but it was fast. I also noticed some MM crontab maintenance jobs did trigger an OOM killer, so I added a bit more RAM (4GB->5GB). Another thing to consider is: we also filter outgoing mails, which is nice to avoid being blacklisted if some bad guy subscribe and post spammy content. But it seems postfix is applying a recipient concurrency of one when sending to the filter, which means if the list has x subscribers then it is checked x times, which is silly. There is no such thing in the configuration so I'm looking into it. This probably affects other installations but this one being heavily loaded… \_o<