Do not create bundles of Omicron zones automatically (#9165)

- Remove automatic zone bundle creation for Omicron zones when the
sled-agent config reconciler shuts down a zone.
- Fixes #9164 in the simplest possible way. Does not actually remove the
zone-bundle functionality entirely or automatic creation for Propolis
zones. Bundles can still be made on-demand.

Author: bnaecker

sled-agent/config-reconciler/src/reconciler_task/zones.rs

@@ -28,7 +28,6 @@ use nexus_sled_agent_shared::inventory::OmicronZoneType;
 use ntp_admin_client::types::TimeSync;
 use omicron_common::address::Ipv6Subnet;
 use omicron_uuid_kinds::OmicronZoneUuid;
-use sled_agent_types::zone_bundle::ZoneBundleCause;
 use sled_agent_types::zone_images::MupdateOverrideReadError;
 use sled_agent_types::zone_images::OmicronZoneImageLocation;
 use sled_agent_types::zone_images::PreparedOmicronZone;
@@ -607,23 +606,6 @@ impl OmicronZone {
         match &self.state {
             ZoneState::Running { running_zone, location: _ } => {
                 info!(log, "shutting down running zone");
-
-                // We only try once to create a zone bundle; if this fails we
-                // move on to the rest of the shutdown process.
-                if let Err(err) = sled_agent_facilities
-                    .zone_bundle_create(
-                        running_zone,
-                        ZoneBundleCause::UnexpectedZone,
-                    )
-                    .await
-                {
-                    warn!(
-                        log,
-                        "Failed to take bundle of zone we're shutting down";
-                        InlineErrorChain::new(err.as_ref()),
-                    );
-                }
-
                 self.resume_shutdown_from_untrack_metrics(
                     sled_agent_facilities,
                     zone_facilities,
@@ -1592,14 +1574,6 @@ mod tests {
         ) {
             self.inner.lock().unwrap().removed_ddm_prefixes.insert(prefix);
         }
-
-        async fn zone_bundle_create(
-            &self,
-            _zone: &RunningZone,
-            _cause: ZoneBundleCause,
-        ) -> anyhow::Result<()> {
-            Ok(())
-        }
     }
 
     // All our tests operate on fake in-memory disks, so the mount config

sled-agent/config-reconciler/src/sled_agent_facilities.rs

@@ -11,7 +11,6 @@ use illumos_utils::zpool::PathInPool;
 use omicron_common::address::Ipv6Subnet;
 use omicron_common::address::SLED_PREFIX;
 use omicron_uuid_kinds::MupdateOverrideUuid;
-use sled_agent_types::zone_bundle::ZoneBundleCause;
 use sled_agent_types::zone_images::PreparedOmicronZone;
 use sled_agent_types::zone_images::RemoveMupdateOverrideResult;
 use sled_agent_types::zone_images::ResolverStatus;
@@ -61,13 +60,6 @@ pub trait SledAgentFacilities: Send + Sync + 'static {
 
     /// Instruct DDM to stop advertising a prefix.
     fn ddm_remove_internal_dns_prefix(&self, prefix: Ipv6Subnet<SLED_PREFIX>);
-
-    /// Create a zone bundle.
-    fn zone_bundle_create(
-        &self,
-        zone: &RunningZone,
-        cause: ZoneBundleCause,
-    ) -> impl Future<Output = anyhow::Result<()>> + Send;
 }
 
 pub trait SledAgentArtifactStore: Send + Sync + 'static {

sled-agent/src/sled_agent.rs

@@ -20,8 +20,8 @@ use crate::services::{self, ServiceManager, UnderlayInfo};
 use crate::support_bundle::logs::SupportBundleLogs;
 use crate::support_bundle::storage::SupportBundleManager;
 use crate::vmm_reservoir::{ReservoirMode, VmmReservoirManager};
+use crate::zone_bundle;
 use crate::zone_bundle::BundleError;
-use crate::zone_bundle::{self, ZoneBundler};
 use anyhow::anyhow;
 use bootstore::schemes::v0 as bootstore;
 use camino::Utf8PathBuf;
@@ -67,7 +67,7 @@ use sled_agent_types::instance::{
 use sled_agent_types::sled::{BaseboardId, StartSledAgentRequest};
 use sled_agent_types::zone_bundle::{
     BundleUtilization, CleanupContext, CleanupCount, CleanupPeriod,
-    PriorityOrder, StorageLimit, ZoneBundleCause, ZoneBundleMetadata,
+    PriorityOrder, StorageLimit, ZoneBundleMetadata,
 };
 use sled_agent_types::zone_images::{
     PreparedOmicronZone, RemoveMupdateOverrideResult, ResolverStatus,
@@ -611,7 +611,6 @@ impl SledAgent {
                 etherstub_vnic,
                 service_manager: services.clone(),
                 metrics_queue: metrics_manager.request_queue(),
-                zone_bundler: long_running_task_handles.zone_bundler.clone(),
             },
             SledAgentArtifactStoreWrapper(Arc::clone(&artifact_store)),
             config_reconciler_spawn_token,
@@ -1294,7 +1293,6 @@ struct ReconcilerFacilities {
     etherstub_vnic: EtherstubVnic,
     service_manager: ServiceManager,
     metrics_queue: MetricsRequestQueue,
-    zone_bundler: ZoneBundler,
 }
 
 impl SledAgentFacilities for ReconcilerFacilities {
@@ -1355,15 +1353,6 @@ impl SledAgentFacilities for ReconcilerFacilities {
             .ddm_reconciler()
             .remove_internal_dns_subnet(prefix);
     }
-
-    async fn zone_bundle_create(
-        &self,
-        zone: &RunningZone,
-        cause: ZoneBundleCause,
-    ) -> anyhow::Result<()> {
-        self.zone_bundler.create(zone, cause).await?;
-        Ok(())
-    }
 }
 
 // Workaround wrapper for orphan rules.