partial impl of query_trusted_key_servers_first option

65% finished

Signed-off-by: strawberry <strawberry@puppygock.gay>

conduwuit-example.toml

@@ -22,10 +22,14 @@
 # YOU NEED TO EDIT THIS
 #server_name = "your.server.name"
 
-# Servers listed here will be used to gather public keys of other servers.
-# Generally, copying this exactly should be enough. (Currently, conduwuit doesn't
-# support batched key requests, so this list should only contain Synapse
-# servers.) Defaults to `matrix.org`
+# Servers listed here will be used to gather public keys of other servers (notary trusted key servers).
+#
+# The default behaviour for conduwuit is to attempt to query trusted key servers before querying the individual servers.
+# This is done for performance reasons, but if you would like to query individual servers before the notary servers
+# configured below, set to
+#
+# (Currently, conduwuit doesn't support batched key requests, so this list should only contain Synapse servers)
+# Defaults to `matrix.org`
 # trusted_servers = ["matrix.org"]
 
 
@@ -228,6 +232,22 @@ allow_check_for_updates = true
 # Defaults to "🏳️‍⚧️" (trans pride flag)
 #new_user_displayname_suffix = "🏳️‍⚧️"
 
+# Option to control whether conduwuit will query your list of trusted notary key servers (`trusted_servers`) for
+# remote homeserver signing keys it doesn't know *first*, or query the individual servers first before falling back to the trusted
+# key servers.
+#
+# The former/default behaviour makes federated/remote rooms joins generally faster because we're querying a single (or list of) server
+# that we know works, is reasonably fast, and is reliable for just about all the homeserver signing keys in the room. Querying individual
+# servers may take longer depending on the general infrastructure of everyone in there, how many dead servers there are, etc.
+#
+# However, this does create an increased reliance on one single or multiple large entities as `trusted_servers` should generally
+# contain long-term and large servers who know a very large number of homeservers.
+#
+# If you don't know what any of this means, leave this and `trusted_servers` alone to their defaults.
+#
+# Defaults to true as this is the fastest option for federation.
+#query_trusted_key_servers_first = true
+
 # Set this to any float value to multiply conduwuit's in-memory LRU caches with.
 # May be useful if you have significant memory to spare to increase performance.
 # Defaults to 1.0.

src/config/mod.rs

@@ -92,6 +92,8 @@ pub struct Config {
     pub jwt_secret: Option<String>,
     #[serde(default = "default_trusted_servers")]
     pub trusted_servers: Vec<OwnedServerName>,
+    #[serde(default = "true_fn")]
+    pub query_trusted_key_servers_first: bool,
     #[serde(default = "default_log")]
     pub log: String,
     #[serde(default)]
@@ -319,6 +321,10 @@ impl fmt::Display for Config {
                 }
                 &lst.join(", ")
             }),
+            (
+                "Query Trusted Key Servers First",
+                &self.query_trusted_key_servers_first.to_string(),
+            ),
             (
                 "TURN username",
                 if self.turn_username.is_empty() {

src/service/globals/mod.rs

@@ -1,30 +1,3 @@
-mod data;
-use argon2::Argon2;
-pub use data::Data;
-use regex::RegexSet;
-use ruma::{
-    serde::Base64, OwnedDeviceId, OwnedEventId, OwnedRoomId, OwnedServerName,
-    OwnedServerSigningKeyId, OwnedUserId,
-};
-
-use sha2::Digest;
-
-use crate::api::server_server::FedDest;
-
-use crate::{services, Config, Error, Result};
-use futures_util::FutureExt;
-use hyper::{
-    client::connect::dns::{GaiResolver, Name},
-    service::Service as HyperService,
-};
-use reqwest::dns::{Addrs, Resolve, Resolving};
-use ruma::{
-    api::{
-        client::sync::sync_events,
-        federation::discovery::{ServerSigningKeys, VerifyKey},
-    },
-    DeviceId, RoomVersionId, ServerName, UserId,
-};
 use std::{
     collections::{BTreeMap, HashMap},
     error::Error as StdError,
@@ -39,11 +12,38 @@ use std::{
     },
     time::{Duration, Instant},
 };
+
+use argon2::Argon2;
+use base64::{engine::general_purpose, Engine as _};
+use futures_util::FutureExt;
+use hyper::{
+    client::connect::dns::{GaiResolver, Name},
+    service::Service as HyperService,
+};
+use regex::RegexSet;
+use reqwest::dns::{Addrs, Resolve, Resolving};
+use ruma::{
+    api::{
+        client::sync::sync_events,
+        federation::discovery::{ServerSigningKeys, VerifyKey},
+    },
+    DeviceId, RoomVersionId, ServerName, UserId,
+};
+use ruma::{
+    serde::Base64, OwnedDeviceId, OwnedEventId, OwnedRoomId, OwnedServerName,
+    OwnedServerSigningKeyId, OwnedUserId,
+};
+use sha2::Digest;
 use tokio::sync::{broadcast, watch::Receiver, Mutex as TokioMutex, Semaphore};
 use tracing::{error, info};
 use trust_dns_resolver::TokioAsyncResolver;
 
-use base64::{engine::general_purpose, Engine as _};
+pub use data::Data;
+
+use crate::api::server_server::FedDest;
+use crate::{services, Config, Error, Result};
+
+mod data;
 
 type WellKnownMap = HashMap<OwnedServerName, (FedDest, String)>;
 type TlsNameMap = HashMap<String, (Vec<IpAddr>, u16)>;
@@ -363,6 +363,10 @@ impl Service<'_> {
         &self.config.trusted_servers
     }
 
+    pub fn query_trusted_key_servers_first(&self) -> bool {
+        self.config.query_trusted_key_servers_first
+    }
+
     pub fn dns_resolver(&self) -> &TokioAsyncResolver {
         &self.dns_resolver
     }

src/service/rooms/event_handler/mod.rs

@@ -1544,39 +1544,11 @@ impl Service {
         Ok(())
     }
 
-    pub(crate) async fn fetch_join_signing_keys(
+    async fn batch_request_signing_keys(
         &self,
-        event: &create_join_event::v2::Response,
-        room_version: &RoomVersionId,
+        mut servers: BTreeMap<OwnedServerName, BTreeMap<OwnedServerSigningKeyId, QueryCriteria>>,
         pub_key_map: &RwLock<BTreeMap<String, BTreeMap<String, Base64>>>,
     ) -> Result<()> {
-        let mut servers: BTreeMap<
-            OwnedServerName,
-            BTreeMap<OwnedServerSigningKeyId, QueryCriteria>,
-        > = BTreeMap::new();
-
-        {
-            let mut pkm = pub_key_map
-                .write()
-                .map_err(|_| Error::bad_database("RwLock is poisoned."))?;
-
-            // Try to fetch keys, failure is okay
-            // Servers we couldn't find in the cache will be added to `servers`
-            for pdu in &event.room_state.state {
-                let _ = self.get_server_keys_from_cache(pdu, &mut servers, room_version, &mut pkm);
-            }
-            for pdu in &event.room_state.auth_chain {
-                let _ = self.get_server_keys_from_cache(pdu, &mut servers, room_version, &mut pkm);
-            }
-
-            drop(pkm);
-        };
-
-        if servers.is_empty() {
-            info!("server is empty, we had all keys locally, not fetching any keys");
-            return Ok(());
-        }
-
         for server in services().globals.trusted_servers() {
             info!("Asking batch signing keys from trusted server {}", server);
             if let Ok(keys) = services()
@@ -1619,13 +1591,16 @@ impl Service {
                     pkm.insert(k.server_name.to_string(), result);
                 }
             }
-
-            if servers.is_empty() {
-                info!("Trusted server supplied all signing keys, no more keys to fetch");
-                return Ok(());
-            }
         }
 
+        Ok(())
+    }
+
+    async fn request_signing_keys(
+        &self,
+        servers: BTreeMap<OwnedServerName, BTreeMap<OwnedServerSigningKeyId, QueryCriteria>>,
+        pub_key_map: &RwLock<BTreeMap<String, BTreeMap<String, Base64>>>,
+    ) -> Result<()> {
         info!("Asking individual servers for signing keys: {servers:?}");
         let mut futures: FuturesUnordered<_> = servers
             .into_keys()
@@ -1641,7 +1616,7 @@ impl Service {
             .collect();
 
         while let Some(result) = futures.next().await {
-            info!("Received new result");
+            debug!("Received new Future result");
             if let (Ok(get_keys_response), origin) = result {
                 info!("Result is from {origin}");
                 if let Ok(key) = get_keys_response.server_key.deserialize() {
@@ -1657,11 +1632,83 @@ impl Service {
                         .insert(origin.to_string(), result);
                 }
             }
-            info!("Done handling result");
+            debug!("Done handling Future result");
+        }
+
+        Ok(())
+    }
+
+    pub(crate) async fn fetch_join_signing_keys(
+        &self,
+        event: &create_join_event::v2::Response,
+        room_version: &RoomVersionId,
+        pub_key_map: &RwLock<BTreeMap<String, BTreeMap<String, Base64>>>,
+    ) -> Result<()> {
+        let mut servers: BTreeMap<
+            OwnedServerName,
+            BTreeMap<OwnedServerSigningKeyId, QueryCriteria>,
+        > = BTreeMap::new();
+
+        {
+            let mut pkm = pub_key_map
+                .write()
+                .map_err(|_| Error::bad_database("RwLock is poisoned."))?;
+
+            // Try to fetch keys, failure is okay
+            // Servers we couldn't find in the cache will be added to `servers`
+            for pdu in &event.room_state.state {
+                let _ = self.get_server_keys_from_cache(pdu, &mut servers, room_version, &mut pkm);
+            }
+            for pdu in &event.room_state.auth_chain {
+                let _ = self.get_server_keys_from_cache(pdu, &mut servers, room_version, &mut pkm);
+            }
+
+            drop(pkm);
+        };
+
+        if servers.is_empty() {
+            info!("We had all keys cached locally, not fetching any keys from remote servers");
+            return Ok(());
+        }
+
+        if services().globals.query_trusted_key_servers_first() {
+            info!("query_trusted_key_servers_first is set to true, querying notary trusted key servers first for homeserver signing keys.");
+
+            self.batch_request_signing_keys(servers.clone(), pub_key_map)
+                .await?;
+
+            if servers.is_empty() {
+                info!("Trusted server supplied all signing keys, no more keys to fetch");
+                return Ok(());
+            }
+
+            info!("Remaining servers left that the notary/trusted servers did not provide: {servers:?}");
+
+            self.request_signing_keys(servers.clone(), pub_key_map)
+                .await?;
+        } else {
+            info!("query_trusted_key_servers_first is set to false, querying individual homeservers first");
+
+            self.request_signing_keys(servers.clone(), pub_key_map)
+                .await?;
+
+            if servers.is_empty() {
+                info!("Individual homeservers supplied all signing keys, no more keys to fetch");
+                return Ok(());
+            }
+
+            info!("Remaining servers left the individual homeservers did not provide: {servers:?}");
+
+            self.batch_request_signing_keys(servers.clone(), pub_key_map)
+                .await?;
         }
 
         info!("Search for signing keys done");
 
+        /*if servers.is_empty() {
+            warn!("Failed to find homeserver signing keys for the remaining servers: {servers:?}");
+        }*/
+
         Ok(())
     }