QID 982386

**Denial of Service 1**

Tendermint 0.33.2 and earlier does not limit the number of P2P connection requests. For each p2p connection, Tendermint allocates XXX bytes. Even though this memory is garbage collected once the connection is terminated (due to duplicate IP or reaching a maximum number of inbound peers), temporary memory spikes can lead to OOM (Out-Of-Memory) exceptions.

Tendermint 0.33.3 (and 0.32.10) limits the total number of P2P incoming connection requests to to `p2p.max_num_inbound_peers + len(p2p.unconditional_peer_ids)`.

Notes:

- Tendermint does not rate limit P2P connection requests per IP (an attacker can saturate all the inbound slots);
- Tendermint does not rate limit HTTP(S) requests. If you expose any RPC endpoints to the public, please make sure to put in place some protection (https://www.nginx.com/blog/rate-limiting-nginx/). We may implement this in the future (https://github.com/tendermint/tendermint/issues/1696).

**Denial of Service 2**

Tendermint 0.33.2 and earlier does not reclaim `activeID` of a peer after it's removed in `Mempool` reactor. This does not happen all the time. It only happens when a connection fails (for any reason) before the `Peer` is created and added to all reactors. `RemovePeer` is therefore called before `AddPeer`, which leads to always growing memory (`activeIDs` map). The `activeIDs` map has a maximum size of 65535 and the node will panic if this map reaches the maximum. An attacker can create a lot of connection attempts (exploiting Denial of Service 1), which ultimately will lead to the node panicking.

Tendermint 0.33.3 (and 0.32.10) claims `activeID` for a peer in `InitPeer`, which is executed before `MConnection` is started.

Notes:

- `InitPeer` function was added to all reactors to combat a similar issue - https://github.com/tendermint/tendermint/issues/3338;
- Denial of Service 2 is independent of Denial of Service 1 and can be executed without it.

- All full nodes (except for validators who are behind closed networks)
- Node's memory usage increases, then it panics either in the mempool or due to OOM.