Support bundle: automatic deletion to maintain free dataset buffer

Author: smklein

dev-tools/omdb/src/bin/omdb/nexus.rs

@@ -70,6 +70,7 @@ use nexus_types::internal_api::background::RegionSnapshotReplacementStartStatus;
 use nexus_types::internal_api::background::RegionSnapshotReplacementStepStatus;
 use nexus_types::internal_api::background::SitrepGcStatus;
 use nexus_types::internal_api::background::SitrepLoadStatus;
+use nexus_types::internal_api::background::SupportBundleAutoDeletionReport;
 use nexus_types::internal_api::background::SupportBundleCleanupReport;
 use nexus_types::internal_api::background::SupportBundleCollectionReport;
 use nexus_types::internal_api::background::SupportBundleCollectionStepStatus;
@@ -2557,6 +2558,7 @@ fn print_task_service_firewall_rule_propagation(details: &serde_json::Value) {
 fn print_task_support_bundle_collector(details: &serde_json::Value) {
     #[derive(Deserialize)]
     struct SupportBundleCollectionStatus {
+        auto_deletion_report: Option<SupportBundleAutoDeletionReport>,
         cleanup_report: Option<SupportBundleCleanupReport>,
         cleanup_err: Option<String>,
         collection_report: Option<SupportBundleCollectionReport>,
@@ -2571,11 +2573,36 @@ fn print_task_support_bundle_collector(details: &serde_json::Value) {
             error, details
         ),
         Ok(SupportBundleCollectionStatus {
+            auto_deletion_report,
             cleanup_report,
             cleanup_err,
             collection_report,
             collection_err,
         }) => {
+            // Print auto-deletion report first (since it runs first)
+            if let Some(SupportBundleAutoDeletionReport {
+                bundles_marked_for_deletion,
+                free_datasets,
+                total_datasets,
+                active_bundles,
+                errors,
+            }) = auto_deletion_report
+            {
+                println!("    Support Bundle Auto-Deletion Report:");
+                println!("      Total debug datasets: {total_datasets}");
+                println!("      Active bundles: {active_bundles}");
+                println!("      Free datasets: {free_datasets}");
+                println!(
+                    "      Bundles marked for deletion: {bundles_marked_for_deletion}"
+                );
+                if !errors.is_empty() {
+                    println!("      Errors:");
+                    for error in errors {
+                        println!("        {error}");
+                    }
+                }
+            }
+
             if let Some(cleanup_err) = cleanup_err {
                 println!("    failed to perform cleanup: {cleanup_err}");
             }

nexus-config/src/nexus_config.rs

@@ -30,6 +30,7 @@ use std::collections::HashMap;
 use std::fmt;
 use std::net::IpAddr;
 use std::net::SocketAddr;
+use std::num::NonZeroU32;
 use std::time::Duration;
 use uuid::Uuid;
 
@@ -489,6 +490,25 @@ pub struct SupportBundleCollectorConfig {
     /// Default: Off
     #[serde(default)]
     pub disable: bool,
+
+    /// Target number of free debug datasets to maintain for new allocations.
+    ///
+    /// When the number of free debug datasets drops below this value, the
+    /// oldest support bundles will be automatically deleted to free up space.
+    ///
+    /// Default: None (auto-deletion disabled)
+    #[serde(default)]
+    pub target_free_datasets: Option<NonZeroU32>,
+
+    /// Minimum number of bundles to retain, even if `target_free_datasets`
+    /// would otherwise trigger deletion.
+    ///
+    /// This prevents aggressive cleanup on small systems where the number of
+    /// debug datasets is less than or equal to `target_free_datasets`.
+    ///
+    /// Default: None (no minimum, can delete all bundles)
+    #[serde(default)]
+    pub min_bundles_to_keep: Option<NonZeroU32>,
 }
 
 #[serde_as]
@@ -1407,6 +1427,8 @@ mod test {
                             SupportBundleCollectorConfig {
                                 period_secs: Duration::from_secs(30),
                                 disable: false,
+                                target_free_datasets: None,
+                                min_bundles_to_keep: None,
                             },
                         physical_disk_adoption: PhysicalDiskAdoptionConfig {
                             period_secs: Duration::from_secs(30),

nexus/db-model/src/schema_versions.rs

@@ -16,7 +16,7 @@ use std::{collections::BTreeMap, sync::LazyLock};
 ///
 /// This must be updated when you change the database schema.  Refer to
 /// schema/crdb/README.adoc in the root of this repository for details.
-pub const SCHEMA_VERSION: Version = Version::new(220, 0, 0);
+pub const SCHEMA_VERSION: Version = Version::new(221, 0, 0);
 
 /// List of all past database schema versions, in *reverse* order
 ///
@@ -28,6 +28,7 @@ static KNOWN_VERSIONS: LazyLock<Vec<KnownVersion>> = LazyLock::new(|| {
         // |  leaving the first copy as an example for the next person.
         // v
         // KnownVersion::new(next_int, "unique-dirname-with-the-sql-files"),
+        KnownVersion::new(221, "bundle-state-index"),
         KnownVersion::new(220, "multicast-implicit-lifecycle"),
         KnownVersion::new(219, "blueprint-sled-last-used-ip"),
         KnownVersion::new(218, "measurements"),