Currently we have a global parameter that allow us to enable/disable outgoing traffic processing
process_outgoing_traffic = on/off
unfortunately its not possible to have graphite data for outgoing traffic while not taking any ban actions in case outgoing gets above threshold.
in my case I have a wrapper (net healer) that gets FNM bans as inputs to take decisions. my wrapper ignores all the outgoing bans that FNM spits.
there was a situation here a layer7 attack (HTTP POST) made my /32 prefix outgoing bandwidth increase above threshold, so FNM banned (outgoing) and my wrapper ignored.
Note: I always use unban timer = 30 seconds.
so I kept receiving outgoing BAN feeds from FNM (and ignoring them) -- because outgoing as above the threshold.
a flag that allow us to disable ban for outgoing (while still process traffic) will help us a lot here, so we can still have graphite history data, but don't take any ban action in case outgoing is above threshold.
such as:
Enable/Disable any actions in case of attack (options: on/off/incoming)
Currently we have a global parameter that allow us to enable/disable outgoing traffic processing
process_outgoing_traffic = on/off
unfortunately its not possible to have graphite data for outgoing traffic while not taking any ban actions in case outgoing gets above threshold.
in my case I have a wrapper (net healer) that gets FNM bans as inputs to take decisions. my wrapper ignores all the outgoing bans that FNM spits. there was a situation here a layer7 attack (HTTP POST) made my /32 prefix outgoing bandwidth increase above threshold, so FNM banned (outgoing) and my wrapper ignored.
Note: I always use unban timer = 30 seconds.
so I kept receiving outgoing BAN feeds from FNM (and ignoring them) -- because outgoing as above the threshold.
a flag that allow us to disable ban for outgoing (while still process traffic) will help us a lot here, so we can still have graphite history data, but don't take any ban action in case outgoing is above threshold.
such as:
Enable/Disable any actions in case of attack (options: on/off/incoming)
enable_ban = incoming