refresh SRV records and send out ServiceRemoved for expired SRV

.github/workflows/build.yml

@@ -27,6 +27,6 @@ jobs:
       - name: Build
         run: cargo build
       - name: Run clippy and fail if any warnings
-        run: cargo clippy -- -W clippy::all -D warnings
+        run: cargo clippy -- -D warnings
       - name: Run tests
         run: cargo test

src/dns_parser.rs

@@ -103,8 +103,14 @@ pub struct DnsRecord {
 impl DnsRecord {
     fn new(name: &str, ty: u16, class: u16, ttl: u32) -> Self {
         let created = current_time_millis();
+
+        // From RFC 6762 section 5.2:
+        // "... The querier should plan to issue a query at 80% of the record
+        // lifetime, and then if no answer is received, at 85%, 90%, and 95%."
         let refresh = get_expiration_time(created, ttl, 80);
+
         let expires = get_expiration_time(created, ttl, 100);
+
         Self {
             entry: DnsEntry::new(name.to_string(), ty, class),
             ttl,
@@ -136,6 +142,36 @@ impl DnsRecord {
         self.refresh = get_expiration_time(self.created, self.ttl, 100);
     }
 
+    /// Returns if this record is due for refresh. If yes, `refresh` time is updated.
+    pub(crate) fn refresh_maybe(&mut self, now: u64) -> bool {
+        if self.is_expired(now) || !self.refresh_due(now) {
+            return false;
+        }
+
+        debug!(
+            "{} qtype {} is due to refresh",
+            &self.entry.name, self.entry.ty
+        );
+
+        // From RFC 6762 section 5.2:
+        // "... The querier should plan to issue a query at 80% of the record
+        // lifetime, and then if no answer is received, at 85%, 90%, and 95%."
+        //
+        // If the answer is received in time, 'refresh' will be reset outside
+        // this function, back to 80% of the new TTL.
+        if self.refresh == get_expiration_time(self.created, self.ttl, 80) {
+            self.refresh = get_expiration_time(self.created, self.ttl, 85);
+        } else if self.refresh == get_expiration_time(self.created, self.ttl, 85) {
+            self.refresh = get_expiration_time(self.created, self.ttl, 90);
+        } else if self.refresh == get_expiration_time(self.created, self.ttl, 90) {
+            self.refresh = get_expiration_time(self.created, self.ttl, 95);
+        } else {
+            self.refresh_no_more();
+        }
+
+        true
+    }
+
     /// Returns the remaining TTL in seconds
     fn get_remaining_ttl(&self, now: u64) -> u32 {
         let remaining_millis = get_expiration_time(self.created, self.ttl, 100) - now;
@@ -190,6 +226,8 @@ pub trait DnsRecordExt: fmt::Debug {
         self.get_record().entry.ty
     }
 
+    /// Resets TTL using `other` record.
+    /// `self.refresh` and `self.expires` are also reset.
     fn reset_ttl(&mut self, other: &dyn DnsRecordExt) {
         self.get_record_mut().reset_ttl(other.get_record());
     }
@@ -1084,7 +1122,16 @@ impl DnsIncoming {
 
             let ty = u16_from_be_slice(&slice[..2]);
             let class = u16_from_be_slice(&slice[2..4]);
-            let ttl = u32_from_be_slice(&slice[4..8]);
+            let mut ttl = u32_from_be_slice(&slice[4..8]);
+            if ttl == 0 && self.is_response() {
+                // RFC 6762 section 10.1:
+                // "...Queriers receiving a Multicast DNS response with a TTL of zero SHOULD
+                // NOT immediately delete the record from the cache, but instead record
+                // a TTL of 1 and then delete the record one second later."
+                // See https://datatracker.ietf.org/doc/html/rfc6762#section-10.1
+
+                ttl = 1;
+            }
             let length = u16_from_be_slice(&slice[8..10]) as usize;
             self.offset += 10;
             let next_offset = self.offset + length;
@@ -1354,19 +1401,22 @@ const fn u32_from_be_slice(s: &[u8]) -> u32 {
     u32::from_be_bytes(u8_array)
 }
 
-/// Returns the time in millis at which this record will have expired
+/// Returns the UNIX time in millis at which this record will have expired
 /// by a certain percentage.
 const fn get_expiration_time(created: u64, ttl: u32, percent: u32) -> u64 {
+    // 'created' is in millis, 'ttl' is in seconds, hence:
+    // ttl * 1000 * (percent / 100) => ttl * percent * 10
     created + (ttl * percent * 10) as u64
 }
 
 #[cfg(test)]
 mod tests {
-    use crate::dns_parser::{TYPE_A, TYPE_AAAA};
+    use crate::dns_parser::get_expiration_time;
 
     use super::{
-        DnsIncoming, DnsNSec, DnsOutgoing, DnsSrv, CLASS_CACHE_FLUSH, CLASS_IN, FLAGS_QR_QUERY,
-        FLAGS_QR_RESPONSE, TYPE_PTR,
+        current_time_millis, DnsIncoming, DnsNSec, DnsOutgoing, DnsRecordExt, DnsSrv,
+        CLASS_CACHE_FLUSH, CLASS_IN, FLAGS_QR_QUERY, FLAGS_QR_RESPONSE, TYPE_A, TYPE_AAAA,
+        TYPE_PTR,
     };
 
     #[test]
@@ -1453,4 +1503,31 @@ mod tests {
         assert_eq!(absent_types[0], TYPE_A);
         assert_eq!(absent_types[1], TYPE_AAAA);
     }
+
+    #[test]
+    fn test_refresh_maybe() {
+        let name = "test_refresh._udp.local.";
+        let ttl = 2;
+        let hostname = "instance1.local.";
+        let mut srv = DnsSrv::new(name, CLASS_IN, ttl, 0, 0, 0, hostname.to_string());
+
+        // refresh is not due yet.
+        let now = current_time_millis();
+        let refreshed = srv.get_record_mut().refresh_maybe(now);
+        assert!(!refreshed);
+
+        // sleep for 80 percent of TTL in millis to reach "refresh" time.
+        let sleep_in_mills = (ttl * 80 * 10) as u64;
+        std::thread::sleep(std::time::Duration::from_millis(sleep_in_mills));
+
+        // refresh is due.
+        let now = current_time_millis();
+        let refreshed = srv.get_record_mut().refresh_maybe(now);
+        assert!(refreshed);
+
+        // refresh time is updated.
+        let dns_record = srv.get_record();
+        let new_refresh = get_expiration_time(dns_record.get_created(), dns_record.ttl, 85);
+        assert_eq!(new_refresh, dns_record.get_refresh_time());
+    }
 }