I'm currently trying out this setup: - logserver at home, connected via ADSL with a fixed IPv4 subnet - IP address: 192.168.0.1 - syslog-ng listening on 127.0.0.1:514 TCP - stunnel listening on 192.168.0.1:514 TCP and connecting to 127.0.0.1:514 TCP - logclient at a remote datacenter - IP address: 172.16.0.1 - syslog-ng delivering to 127.0.0.1:514 TCP - stunnel listening on 127.0.0.1:514 TCP and connecting to 192.168.0.1:514 TCP Here are the STunnel configs: --8<--[ stunnel.conf on logserver (excluding cert/key stuff) ]--8<-- [syslog-ng] client = no accept = 192.168.0.1:514 local = 127.0.0.1 connect = 127.0.0.1:514 --8<-- --8<--[ stunnel.conf on logclient (excluding cert/key stuff) ]--8<-- [syslog-ng] client = yes accept = 127.0.0.1:514 local = 172.16.0.1 connect = 192.168.0.1:514 --8<-- This works fine so far, logs are sent over the encrypted tunnel without problems. Except one situation: the STunnel or Syslog-NG itself on the logserver goes down. Then this happens on the logclient: --8<-- syslog-ng[30119]: EOF occurred while idle; fd='33' --8<-- There are *lots* of those messages, like 20 per second. Is it possible to prevent Syslog-NG of creating such messages? -- Wolfram Schlich