Tor developer David Goulet announced 0.4.9.17 as an emergency response to an ongoing distributed denial-of-service attack. The November 13 announcement says a flood of client circuits has overwhelmed relays and affected most relay operators. Circuits are the paths Tor builds through its relay network.

The official release notes identify TROVE-2026-031, an IPv6 handling flaw in the denial-of-service defenses. The fix tracks clients by their /64 network prefix, a group of IPv6 addresses, rather than treating every address as a separate client. That gives the defense a shared accounting boundary across addresses in the same group.

The notes trace the bug to 0.3.3.2-alpha and provide the fix in 0.4.9.17. That origin does not mean obsolete Tor branches are supported choices. Relay operators should use a currently supported, patched release through their normal trusted distribution.

Update the relay, then check its operation

Tor asks relay operators to upgrade as soon as possible and says no additional action is required after upgrading. Confirm that the running relay actually uses the updated package or container. Watch its resource use and service health after the change, especially if it shares a connection with other services.

Evidence status: an ongoing attack reported by the network’s developers, with a released defensive fix. The announcement says the team is evaluating further defenses and expects additional releases. It does not claim that this update ends the attack.

The release also removes the bastet directory authority’s keys because of a potential compromise. That is a precaution described in the notes, not confirmation that the authority was compromised or responsible for the flooding.

What the thread adds

In the Hacker News discussion, Actor describes a home relay overloading a shared internet connection. They report that stopping its host restored normal speeds and that lowering its configured bandwidth had not resolved their problem. This is one operator’s account, not a measured network-wide effect or proof that the new release fails.

That experience makes the shared connection worth checking alongside the relay process itself. Follow Tor’s upgrade guidance rather than treating a reader’s decision to stop operating a guard relay as a general prescription. Other comments speculate about attacker identity; the release announcement supplies no attribution.

From a 55-comment thread, in HN’s own order.

Caveat: this is an availability incident. The reviewed sources do not demonstrate that the flooding exposed users’ browsing or defeated Tor’s anonymity protections.