fix: Ignore extra HealthStatus fields in service_monitor

Signed-off-by: Marcel Guzik <[email protected]>

crates/core/tedge_api/src/health.rs

@@ -10,7 +10,6 @@ use mqtt_channel::Topic;
 use serde::Deserialize;
 use serde::Serialize;
 use serde_json::json;
-use serde_json::Value as JsonValue;
 use std::fmt::Display;
 use std::process;
 use std::sync::Arc;
@@ -97,11 +96,16 @@ impl ServiceHealthTopic {
     }
 }
 
+/// Payload of the health status message.
+///
+/// Contains only fields required for the payload to be considered a valid health status message.
+/// Other components are free to require additional fields for their purposes.
+///
+/// https://thin-edge.github.io/thin-edge.io/operate/troubleshooting/monitoring-service-health/
 #[derive(Deserialize, Serialize, Debug, Default)]
 pub struct HealthStatus {
+    /// Current status of the service, synced by the mapper to the cloud
     pub status: Status,
-    pub pid: Option<u32>,
-    pub time: Option<JsonValue>,
 }
 
 #[derive(Clone, Debug, PartialEq, Eq, Deserialize, Serialize)]
@@ -146,11 +150,7 @@ impl HealthStatus {
                     Ok("0") => Status::Down,
                     _ => Status::default(),
                 };
-                HealthStatus {
-                    status,
-                    pid: None,
-                    time: None,
-                }
+                HealthStatus { status }
             } else {
                 serde_json::from_slice(message.payload()).unwrap_or_default()
             };

crates/core/tedge_watchdog/src/systemd_watchdog.rs

@@ -1,4 +1,3 @@
-use crate::error::WatchdogError;
 use anyhow::Context;
 use freedesktop_entry_parser::parse_entry;
 use futures::channel::mpsc;
@@ -8,6 +7,9 @@ use futures::StreamExt;
 use mqtt_channel::MqttMessage;
 use mqtt_channel::PubChannel;
 use mqtt_channel::Topic;
+use serde::Deserialize;
+use serde::Serialize;
+use serde_json::Value as JsonValue;
 use std::path::PathBuf;
 use std::process;
 use std::process::Command;
@@ -20,7 +22,6 @@ use tedge_api::mqtt_topics::Channel;
 use tedge_api::mqtt_topics::EntityTopicId;
 use tedge_api::mqtt_topics::MqttSchema;
 use tedge_api::mqtt_topics::OperationType;
-use tedge_api::HealthStatus;
 use tedge_config::TEdgeConfigLocation;
 use tedge_utils::timestamp::IsoOrUnix;
 use time::OffsetDateTime;
@@ -29,6 +30,8 @@ use tracing::error;
 use tracing::info;
 use tracing::warn;
 
+use crate::error::WatchdogError;
+
 const SERVICE_NAME: &str = "tedge-watchdog";
 
 /// How many times more often do we send notify to systemd watchdog, than is necessary from the
@@ -38,6 +41,16 @@ const SERVICE_NAME: &str = "tedge-watchdog";
 /// a timing misalignment.
 const NOTIFY_SEND_FREQ_RATIO: u64 = 4;
 
+/// A subset of fields of health status payload required by the watchdog.
+///
+/// https://thin-edge.github.io/thin-edge.io/operate/troubleshooting/monitoring-service-health/
+#[derive(Debug, Serialize, Deserialize)]
+struct HealthStatusExt {
+    /// Used for tracking service restarts
+    pub pid: Option<u32>,
+    pub time: Option<JsonValue>,
+}
+
 pub async fn start_watchdog(tedge_config_dir: PathBuf) -> Result<(), anyhow::Error> {
     // Send ready notification to systemd.
     notify_systemd(process::id(), "--ready")?;
@@ -128,7 +141,6 @@ async fn start_watchdog_for_tedge_services(tedge_config_dir: PathBuf) {
 
                 let tedge_config_location = tedge_config_location.clone();
                 watchdog_tasks.push(tokio::spawn(async move {
-                    //
                     let interval = Duration::from_secs((interval / NOTIFY_SEND_FREQ_RATIO).max(1));
                     monitor_tedge_service(
                         tedge_config_location,
@@ -243,11 +255,11 @@ async fn monitor_tedge_service(
 async fn get_latest_health_status_message(
     request_timestamp: OffsetDateTime,
     messages: &mut mpsc::UnboundedReceiver<MqttMessage>,
-) -> Result<HealthStatus, WatchdogError> {
+) -> Result<HealthStatusExt, WatchdogError> {
     while let Some(message) = messages.next().await {
         if let Ok(message) = message.payload_str() {
             debug!("Health response received: {message}");
-            if let Ok(health_status) = serde_json::from_str::<HealthStatus>(message) {
+            if let Ok(health_status) = serde_json::from_str::<HealthStatusExt>(message) {
                 if health_status.time.is_none() {
                     error!("Ignoring invalid health response: {health_status:?} without a `time` field in it");
                     continue;

crates/extensions/c8y_mapper_ext/src/service_monitor.rs

@@ -32,11 +32,8 @@ pub fn convert_health_status_message(
         return vec![];
     }
 
-    let HealthStatus {
-        status,
-        pid: _,
-        time: _,
-    } = HealthStatus::try_from_health_status_message(message, mqtt_schema).unwrap();
+    let HealthStatus { status } =
+        HealthStatus::try_from_health_status_message(message, mqtt_schema).unwrap();
 
     let display_name = entity
         .other
@@ -83,6 +80,15 @@ mod tests {
         r#"102,test_device:device:main:service:tedge-mapper-c8y,service,tedge-mapper-c8y,up"#;
         "service-monitoring-thin-edge-device"
     )]
+    // If there are any problems with fields other than `status`, we want to ignore them and still send status update
+    #[test_case(
+        "test_device",
+        "te/device/main/service/tedge-mapper-c8y/status/health",
+        r#"{"unrecognised_field": [42], "time": "invalid timestamp", "pid": "invalid pid", "status": "up"}"#,
+        "c8y/s/us",
+        r#"102,test_device:device:main:service:tedge-mapper-c8y,service,tedge-mapper-c8y,up"#;
+        "service-monitoring-thin-edge-device-optional-fields-invalid"
+    )]
     #[test_case(
         "test_device",
         "te/device/child/service/tedge-mapper-c8y/status/health",