Report what a device fails to deliver to the controller

CAPABILITIES.md

@@ -65,6 +65,8 @@ introduced the value, not stated anywhere authoritative — corrections welcome.
 | `LOC_REBOOT_COLLECT_INFO` = 19 | cfg | LOC-initiated reboot and collect-info; `LOCConfig.datastore_collect_info_id` **(?)** |
 | `SMART_REPORT` = 20 | rpt | S.M.A.R.T. data in `ZHardwareHealth.disks`, superseding the deprecated `ZInfoHardware.disks` |
 | `REPORT_TPM_EVENTLOG` = 21 | rpt | `ZAttestQuote.tpm_binary_event_log`, superseding the deprecated `ZAttestQuote.event_log` |
+| `APP_INSTANCE_NET_INTERFACE_CHANGE` = 22 | cfg | adding or removing an `AppInstanceConfig.interfaces` entry needs only `restart`, not `purge`, so the app keeps its volumes; the `AppInstanceConfig` comment still describes the older purge requirement **(?)** |
+| `DEFERRED_QUEUE_METRICS` = 23 | rpt | `deviceMetric.deferred_queue`, and `urlcloudMetric.retriableErrCount`, `.rejectedErrCount`, `.deliveredMsgCount` |
 
 ### Adding a value
 

proto/info/info.proto

@@ -722,6 +722,7 @@ enum APICapability {
   API_CAPABILITY_SMART_REPORT = 20; // Support for S.M.A.R.T. info on physical storage devices
   API_CAPABILITY_REPORT_TPM_EVENTLOG = 21; // Support for reporting TPM Event Log "as is" without parsing and selectively reporting events
   API_CAPABILITY_APP_INSTANCE_NET_INTERFACE_CHANGE = 22; // Support for adding and removing network interfaces without purge, but restart or the edge app
+  API_CAPABILITY_DEFERRED_QUEUE_METRICS = 23; // Reports deviceMetric.deferred_queue and the per-URL retriable, rejected and delivered counts
   // Append new values as new API features are implemented; never insert or
   // renumber, since a value asserts that all lower values are also supported.
 }

proto/metrics/metrics.proto

@@ -136,6 +136,20 @@ message urlcloudMetric {
   int64 recvByteCount = 7;
   int64 total_time_spent = 8; // Total time spent servicing requests that succeeded
   int64 sess_resume_count = 9;  // Total counts of the TLS session resumption
+
+  int64 retriableErrCount = 10; // Answered with a status which may not repeat, such as a server error or a rate limit; the message is kept and offered again
+  int64 rejectedErrCount = 11;  // Answered with a rejection of the request itself, so the message is given up on
+  int64 deliveredMsgCount = 12; // Answered with a success status, i.e. accepted; narrower than sentMsgCount, which counts any answer
+}
+
+// What the device is holding back or has given up on delivering to the controller
+message deferredQueueMetric {
+  uint32 undeliveredMsgCount = 1;                     // Messages waiting to be delivered
+  google.protobuf.Timestamp oldestUndeliveredMsg = 2; // When the oldest of them was produced; unset when there are none
+  uint64 droppedMsgCount = 3;                         // Messages given up on since boot, i.e. state the controller was never told
+  uint64 rejectedMsgCount = 4;                        // Of those, the ones the controller rejected
+  uint64 supersededMsgCount = 5;                      // Of those, the ones a later periodic publication supersedes
+  google.protobuf.Timestamp lastDroppedMsg = 6;       // When the most recent one was given up on
 }
 
 // Failures and successes for the object decryption
@@ -303,6 +317,9 @@ message deviceMetric {
 
   // Bond adapter metrics
   repeated BondMetrics bond_metrics = 31;
+
+  // Queue of messages waiting to reach the controller; per device, not per port
+  deferredQueueMetric deferred_queue = 32;
 }
 
 message AclMetric {