Hi Antoine After some investigation I found that the count for the mentioned error was not slowly evolving, it was appearing at beginning of March a lot more than before. So I checked for changes on my side and found I had prolonged delays for hosts that are dns blacklisted too much (I set a small penalty for hosts on DNSRBL and a larger delay for positive spamassassin results to waste spammers resources).
Some providers have very short time-outs and it seems TLS connection cannot be left open too long without sending anything anyway, which both leads to the error in my logs. Especially sendgrid has short delays and their hosts are often on blacklists, so I saw the error coming up very often with sendgrid hosts. As soon as I reduced the delay for blacklisted hosts to 15 seconds again instead of 35, the error disappeared and things run normally again now. So the error is mine and I apologize for the noise on the list. Thank you all for your answers. Best regards, Adrian. PS: I know it is bad practice to slow down legitimate connections, but I see it a good practice to slow down spammers. Systems that appear on blacklists do have a reason being listed and so the small penalty is acceptable in my opinion even though the message being received is a legitimate one. On 30.03.17 14:20, Antoine Beaupré wrote: > > So your first task is to contact the maintainer of that backport to make > sure it's updated, with a CC on the backports list. :) >
