How It Works
The command comes from GitHub, where you can read every line. It only reads: it changes no setting, installs nothing, opens no port and deletes its temporary files when it finishes. It takes about 70 seconds.
What It Reads
For 60 seconds it reads the network card’s traffic counters and the kernel’s own TCP, UDP and ICMP counters every two seconds. Then it counts the server’s connections by state, by remote address and by local port, reads the connection-tracking table when run with sudo, and lists the ports that are listening with the name of the program on each.
What It Sends
Only those counts and totals. Never your server’s hostname or its own addresses, any packet contents, file contents or command lines. The addresses that connect to your server are sent so they can be checked against Frabs’ threat feeds.
What You Get
Whether the server is under attack right now, and what kind: SYN floods, UDP floods, reflection attacks through DNS, NTP, Memcached and other services, ICMP floods, connection floods and SSH or Remote Desktop login floods. You also see how much traffic came in, the addresses and networks it came from, and which services and game servers were hit.
A Minute Is a Snapshot
If the server was slow or unreachable earlier, run the check again while it’s happening. Frabs itself watches every server all the time and blocks attacks as they start.