Attack protection
You can see the attacks on your public addresses: when, how big, what kind, and what was done about them.
What you see
We measure the traffic arriving from the internet on each of your public addresses. When it looks like an attack, it shows up in the panel:
- on the server page and next to the address, for as long as it lasts: Under attack, with the peak and what is being done;
- in the Protection section of the Cloud page (Attacks on your addresses): the project's attacks, most recent first, with address, server, start and duration (ongoing if it is still going on), peak in packets and bits per second, type (for example DNS amplification, SYN flood) and what was done. Opening an attack shows the traffic classes, one per row;
- in AI mode, by asking for example "Am I under attack?".
If the attacked address is the project's NAT gateway, the row says so (NAT gateway).
What was done
| In the panel | In the API (action) |
Meaning |
|---|---|---|
| Reaches the server | none |
It reaches the server: its firewall lets that traffic in |
| Dropped by the host | filtered |
The server's firewall does not let it in and the host drops it; the link to the host still carries it, though |
| Limited by the host | rate_limited |
The host rate-limits it (known UDP amplifications, above a threshold) |
| Stopped at the network edge | upstream |
Handled at the edge of the network (filters, scrubbing, blackholing) |
If you see Reaches the server for traffic you do not need, close or narrow the port in the firewall: from then on the host drops it.
For developers
Cloud/getAttackList / cloud_get_attack_list (viewer role, read-only), with state: ongoing (only those in progress) or all (the default), and server_id for a single server. Each attack has address, server_id (null for the NAT gateway), started_at, ended_at (null while it lasts), peak_pps, peak_bps, action and the classes (for each one class, kind, peaks, packets, bytes, dropped, action).