Return different process exit codes from iotedged after configured re…

…try count has been exceeded (#1190)

edgelet/contrib/config/linux/config.yaml

@@ -144,6 +144,33 @@ agent:
 
 hostname: "<ADD HOSTNAME HERE>"
 
+###############################################################################
+# Watchdog settings
+###############################################################################
+#
+# The IoT edge daemon has a watchdog that periodically checks the health of the
+# Edge Agent module and restarts it if it's down.
+#
+# max_retries - Configures the number of retry attempts that the IoT edge daemon
+#               should make for failed operations before failing with a fatal error.
+#
+#               If this configuration is not specified, the daemon keeps retrying
+#               on errors and doesn't fail fatally.
+# 
+#               On a fatal failure, the daemon returns an exit code which
+#               signifies the kind of error encountered. Currently, the following
+#               error codes are returned by the daemon -
+#
+#               150 - Invalid Device ID specified.
+#               151 - Invalid IoT hub configuration.
+#               152 - Invalid SAS token used to call IoT hub.
+#                     This could signal an invalid SAS key.
+#               1 - All other errors.
+###############################################################################
+
+#watchdog:
+#  max_retries: 2
+
 ###############################################################################
 # Connect settings
 ###############################################################################

edgelet/contrib/config/linux/debian/config.yaml

@@ -144,6 +144,33 @@ agent:
 
 hostname: "<ADD HOSTNAME HERE>"
 
+###############################################################################
+# Watchdog settings
+###############################################################################
+#
+# The IoT edge daemon has a watchdog that periodically checks the health of the
+# Edge Agent module and restarts it if it's down.
+#
+# max_retries - Configures the number of retry attempts that the IoT edge daemon
+#               should make for failed operations before failing with a fatal error.
+#
+#               If this configuration is not specified, the daemon keeps retrying
+#               on errors and doesn't fail fatally.
+#
+#               On a fatal failure, the daemon returns an exit code which
+#               signifies the kind of error encountered. Currently, the following
+#               error codes are returned by the daemon -
+#
+#               150 - Invalid Device ID specified.
+#               151 - Invalid IoT hub configuration.
+#               152 - Invalid SAS token used to call IoT hub.
+#                     This could signal an invalid SAS key.
+#               1 - All other errors.
+###############################################################################
+
+#watchdog:
+#  max_retries: 2
+
 ###############################################################################
 # Connect settings
 ###############################################################################

edgelet/contrib/config/windows/config.yaml

@@ -144,6 +144,33 @@ agent:
 
 hostname: "<ADD HOSTNAME HERE>"
 
+###############################################################################
+# Watchdog settings
+###############################################################################
+#
+# The IoT edge daemon has a watchdog that periodically checks the health of the
+# Edge Agent module and restarts it if it's down.
+#
+# max_retries - Configures the number of retry attempts that the IoT edge daemon
+#               should make for failed operations before failing with a fatal error.
+#
+#               If this configuration is not specified, the daemon keeps retrying
+#               on errors and doesn't fail fatally.
+#
+#               On a fatal failure, the daemon returns an exit code which
+#               signifies the kind of error encountered. Currently, the following
+#               error codes are returned by the daemon -
+#
+#               150 - Invalid Device ID specified.
+#               151 - Invalid IoT hub configuration.
+#               152 - Invalid SAS token used to call IoT hub.
+#                     This could signal an invalid SAS key.
+#               1 - All other errors.
+###############################################################################
+
+#watchdog:
+#  max_retries: 2
+
 ###############################################################################
 # Connect settings
 ###############################################################################

edgelet/edgelet-config/src/lib.rs

@@ -25,6 +25,7 @@ use sha2::{Digest, Sha256};
 use url::Url;
 
 use edgelet_core::crypto::MemoryKey;
+use edgelet_core::watchdog::RetryLimit;
 use edgelet_core::ModuleSpec;
 use edgelet_utils::log_failure;
 
@@ -358,6 +359,18 @@ impl Certificates {
     }
 }
 
+#[derive(Debug, Default, Deserialize, Serialize)]
+pub struct WatchdogSettings {
+    #[serde(default)]
+    max_retries: RetryLimit,
+}
+
+impl WatchdogSettings {
+    pub fn max_retries(&self) -> &RetryLimit {
+        &self.max_retries
+    }
+}
+
 #[derive(Debug, Deserialize, Serialize)]
 pub struct Settings<T> {
     provisioning: Provisioning,
@@ -368,6 +381,8 @@ pub struct Settings<T> {
     homedir: PathBuf,
     moby_runtime: MobyRuntime,
     certificates: Option<Certificates>,
+    #[serde(default)]
+    watchdog: WatchdogSettings,
 }
 
 impl<T> Settings<T>
@@ -432,6 +447,10 @@ where
         self.certificates.as_ref()
     }
 
+    pub fn watchdog(&self) -> &WatchdogSettings {
+        &self.watchdog
+    }
+
     pub fn diff_with_cached(&self, path: &Path) -> bool {
         fn diff_with_cached_inner<T>(
             cached_settings: &Settings<T>,
@@ -510,6 +529,7 @@ mod tests {
     use super::*;
     use config::{Config, File, FileFormat};
     use edgelet_docker::DockerConfig;
+    use std::cmp::Ordering;
     use std::fs::File as FsFile;
     use std::io::Write;
     use tempdir::TempDir;
@@ -830,4 +850,14 @@ mod tests {
         let create_options = settings.agent().config().create_options();
         assert_eq!(create_options.hostname(), Some("VAluE3"));
     }
+
+    #[test]
+    fn watchdog_settings_are_read() {
+        let settings = Settings::<DockerConfig>::new(Some(Path::new(GOOD_SETTINGS)));
+        println!("{:?}", settings);
+        assert!(settings.is_ok());
+        let s = settings.unwrap();
+        let watchdog_settings = s.watchdog();
+        assert_eq!(watchdog_settings.max_retries().compare(3), Ordering::Equal);
+    }
 }

edgelet/edgelet-config/test/linux/sample_settings.yaml

@@ -14,6 +14,9 @@ agent:
     auth: {}
 hostname: "localhost"
 
+watchdog:
+  max_retries: 3
+
 # Sets the connection uris for clients
 connect:
   workload_uri: "http://localhost:8081"

edgelet/edgelet-config/test/linux/sample_settings1.yaml

@@ -12,6 +12,9 @@ agent:
     auth: {}
 hostname: "localhost"
 
+watchdog:
+  max_retries: 3
+
 # Sets the connection uris for clients
 connect:
   workload_uri: "http://localhost:8081"

edgelet/edgelet-config/test/linux/sample_settings2.yaml

@@ -14,6 +14,9 @@ agent:
     auth: {}
 hostname: "localhost"
 
+watchdog:
+  max_retries: 3
+
 # Sets the connection uris for clients
 connect:
   workload_uri: "http://localhost:8081"