Unify filesystem storage accounting

.github/workflows/ci.yaml

@@ -190,6 +190,58 @@ jobs:
           artifact-name: worker-executor-tests-${{ matrix.group.name }}-report
           github-token: ${{ secrets.GITHUB_TOKEN }}
 
+  managed-xfs-tests:
+    name: managed-xfs-tests
+    runs-on: blacksmith-8vcpu-ubuntu-2404
+    needs:
+      - build-and-store
+      - merge-test-components
+    timeout-minutes: 45
+    steps:
+      - uses: actions/checkout@v5
+        with:
+          fetch-depth: 1
+      - uses: ./.github/actions/setup-rust
+        with:
+          use-cache: 'false'
+          install-cargo-binstall: 'true'
+      - uses: ./.github/actions/restore-binaries
+        with:
+          run-id: ${{ github.run_id }}
+          fail-on-cache-miss: 'true'
+      - uses: ./.github/actions/restore-test-components
+        with:
+          run-id: ${{ github.run_id }}
+      - name: Set up Lima
+        id: lima
+        env:
+          GH_TOKEN: ${{ github.token }}
+          LIMA_VERSION: v2.2.0
+        run: |
+          sudo apt-get update -qq
+          sudo apt-get install -qqy --no-install-recommends ovmf qemu-system-x86 qemu-utils
+          test -c /dev/kvm
+          sudo chown "$(whoami)" /dev/kvm
+          archive="lima-${LIMA_VERSION#v}-Linux-x86_64.tar.gz"
+          curl -fOSL "https://github.com/lima-vm/lima/releases/download/${LIMA_VERSION}/${archive}"
+          gh attestation verify --owner lima-vm "${archive}"
+          sudo tar -C /usr/local -xzf "${archive}"
+          rm "${archive}"
+          echo "version=${LIMA_VERSION}" >> "${GITHUB_OUTPUT}"
+      - name: Cache Lima images
+        uses: actions/cache@v5
+        with:
+          path: ~/.cache/lima
+          key: lima-${{ runner.os }}-${{ steps.lima.outputs.version }}
+      - name: Run managed XFS suite
+        env:
+          GOLEM_MANAGED_XFS_REUSE_TEST_BINARIES: 1
+          GOLEM_MANAGED_XFS_TARGET_DIR: ${{ github.workspace }}/target
+          GOLEM_MANAGED_XFS_CARGO_TEST_R: ${{ github.workspace }}/target/cargo-test-r
+        run: |
+          cp "$(command -v cargo-test-r)" "${GOLEM_MANAGED_XFS_CARGO_TEST_R}"
+          integration-tests/scripts/managed-filesystem/run-lima.sh
+
   build-cli-test-bins:
     env:
       CARGO_BUILD_JOBS: 20
@@ -1020,6 +1072,7 @@ jobs:
       - unit-tests-and-checks
       - golem-schema-guest
       - worker-tests
+      - managed-xfs-tests
       - it
       - it-cli
     if: "startsWith(github.ref, 'refs/tags/v')"

Cargo.toml

@@ -76,6 +76,7 @@ bitflags = "2.6.0"
 blake3 = { version = "1.8.2", features = ["rayon"] }
 bytes = "1.10.1"
 camino = "1.1.10"
+cap-fs-ext = "3.4.5" # keep in sync with wasmtime
 cap-std = "3.4.5" # keep in sync with wasmtime
 cap-time-ext = "3.4.5"
 cargo_metadata = "0.21.0"
@@ -153,6 +154,8 @@ kube-derive = "0.98.0"
 lazy_static = "1.5.0"
 leb128 = "0.2.5"
 lenient_bool = "0.1.1"
+libc = "0.2.186"
+linux-raw-sys = { version = "0.12.1", features = ["general", "ioctl"] }
 log = "0.4.26"
 mac_address = "1.1.8"
 mappable-rc = "0.1.1"
@@ -220,6 +223,7 @@ rsa = "0.9.7"
 
 rust_decimal = "1.39.0"
 rustc-hash = "2.1.1"
+rustix = { version = "1.1.4", features = ["fs"] }
 rustls = { version = "0.23.23", features = ["ring"] }
 rustls-pemfile = "2.2.0"
 sanitize-filename = "0.6.0"

cli/golem-cli/command-output-schema/command-output.schema.json

@@ -4347,7 +4347,6 @@
         { "$ref": "#/definitions/PublicOplogEntrySuccessfulUpdate" },
         { "$ref": "#/definitions/PublicOplogEntryFailedUpdate" },
         { "$ref": "#/definitions/PublicOplogEntryGrowMemory" },
-        { "$ref": "#/definitions/PublicOplogEntryFilesystemStorageUsageUpdate" },
         { "$ref": "#/definitions/PublicOplogEntryCreateResource" },
         { "$ref": "#/definitions/PublicOplogEntryDropResource" },
         { "$ref": "#/definitions/PublicOplogEntryLog" },
@@ -4553,12 +4552,6 @@
       "properties": { "type": { "const": "GrowMemory" }, "timestamp": { "type": "string" }, "delta": { "type": "integer", "minimum": 0 } },
       "additionalProperties": false
     },
-    "PublicOplogEntryFilesystemStorageUsageUpdate": {
-      "type": "object",
-      "required": ["type", "timestamp", "delta"],
-      "properties": { "type": { "const": "FilesystemStorageUsageUpdate" }, "timestamp": { "type": "string" }, "delta": { "type": "integer" } },
-      "additionalProperties": false
-    },
     "PublicOplogEntryCreateResource": {
       "type": "object",
       "required": ["type", "timestamp", "id", "name", "owner"],

cli/golem-cli/src/model/agent/oplog.rs

@@ -354,17 +354,6 @@ impl TextOutput for PublicOplogEntry {
                     format_id(&format_binary_size(&params.delta)),
                 ));
             }
-            PublicOplogEntry::FilesystemStorageUsageUpdate(params) => {
-                logln(format_message_highlight("STORAGE USAGE UPDATE"));
-                logln(format!(
-                    "{pad}at:                {}",
-                    format_id(&params.timestamp)
-                ));
-                logln(format!(
-                    "{pad}delta:             {}",
-                    format_id(&params.delta.to_string()),
-                ));
-            }
             PublicOplogEntry::CreateResource(params) => {
                 logln(format_message_highlight("CREATE RESOURCE"));
                 logln(format!(

cli/golem-cli/src/model/cli_output/tests.rs

@@ -1972,10 +1972,6 @@ fn sample_public_oplog_entries() -> Vec<golem_common::model::oplog::PublicOplogE
             timestamp: timestamp(),
             delta: 64,
         }),
-        PublicOplogEntry::FilesystemStorageUsageUpdate(FilesystemStorageUsageUpdateParams {
-            timestamp: timestamp(),
-            delta: -5,
-        }),
         PublicOplogEntry::CreateResource(CreateResourceParams {
             timestamp: timestamp(),
             id: AgentResourceId(1),

cli/golem-cli/wit/deps/golem-1.x/golem-oplog.wit

@@ -406,11 +406,6 @@ interface oplog {
         delta: u64
     }
 
-    record filesystem-storage-usage-update-parameters {
-        timestamp: datetime,
-        delta: s64
-    }
-
     type agent-resource-id = u64;
 
     record create-resource-parameters {
@@ -566,8 +561,6 @@ interface oplog {
         exceeded-table-limit,
         exceeded-http-call-limit,
         exceeded-rpc-call-limit,
-        node-out-of-filesystem-storage,
-        agent-exceeded-filesystem-storage-limit,
         agent-terminated-by-quota(agent-terminated-by-quota-error),
         ephemeral-sleep-too-long(ephemeral-sleep-too-long),
         ephemeral-fuel-exhausted(ephemeral-fuel-exhausted),
@@ -786,8 +779,6 @@ interface oplog {
         failed-update(failed-update-parameters),
         /// Increased total linear memory size
         grow-memory(grow-memory-parameters),
-        /// Updated filesystem usage by a signed delta
-        filesystem-storage-usage-update(filesystem-storage-usage-update-parameters),
         /// Created a resource instance
         create-resource(raw-create-resource-parameters),
         /// Dropped a resource instance
@@ -895,8 +886,6 @@ interface oplog {
         failed-update(failed-update-parameters),
         /// Increased total linear memory size
         grow-memory(grow-memory-parameters),
-        /// Updated filesystem usage by a signed delta
-        filesystem-storage-usage-update(filesystem-storage-usage-update-parameters),
         /// Created a resource instance
         create-resource(create-resource-parameters),
         /// Dropped a resource instance

docs/src/content/next/how-to-guides/common/golem-debug-agent-history.mdx

@@ -58,7 +58,6 @@ Each text entry is printed with its index (e.g. `#00042:`) followed by a labeled
 | `INTERRUPTED` / `EXITED` | Agent interrupted or exited |
 | `NOP` | No-operation marker |
 | `JUMP` | Oplog jump — shows from/to indices |
-| `STORAGE USAGE UPDATE` | Filesystem storage usage change |
 
 ### Examples
 

golem-api-grpc/proto/golem/worker/public_oplog.proto

@@ -22,8 +22,8 @@ import "golem/schema/schema.proto";
 // removed and replaced by `Start`, `End`, `Cancelled`. Their tag numbers are reserved to
 // prevent accidental reuse should we ever need rolling-version compatibility in the future.
 message OplogEntry {
-  reserved 2, 14, 15;
-  reserved "HostCall", "BeginRemoteWrite", "EndRemoteWrite";
+  reserved 2, 14, 15, 41;
+  reserved "HostCall", "BeginRemoteWrite", "EndRemoteWrite", "FilesystemStorageUsageUpdate";
 
   oneof entry {
     CreateParameters Create = 1;
@@ -60,7 +60,6 @@ message OplogEntry {
     RemoteTransactionParameters RolledBackRemoteTransaction = 38;
     SnapshotDataParameters Snapshot = 39;
     OplogProcessorCheckpointParameters OplogProcessorCheckpoint = 40;
-    FilesystemStorageUsageUpdateParameters FilesystemStorageUsageUpdate = 41;
     SetRetryPolicyParameters SetRetryPolicy = 42;
     RemoveRetryPolicyParameters RemoveRetryPolicy = 43;
     StartParameters Start = 44;
@@ -411,11 +410,6 @@ message GrowMemoryParameters {
   uint64 delta = 2;
 }
 
-message FilesystemStorageUsageUpdateParameters {
-  google.protobuf.Timestamp timestamp = 1;
-  int64 delta = 2;
-}
-
 message CreateResourceParameters {
   google.protobuf.Timestamp timestamp = 1;
   uint64 resource_id = 2;

golem-api-grpc/proto/golem/worker/raw_oplog.proto

@@ -16,6 +16,9 @@ import "google/protobuf/timestamp.proto";
 
 // An oplog entry in its raw/internal representation.
 message RawOplogEntry {
+  reserved 21;
+  reserved "filesystem_storage_usage_update";
+
   google.protobuf.Timestamp timestamp = 1;
   oneof entry {
     RawCreateParameters create = 2;
@@ -34,7 +37,6 @@ message RawOplogEntry {
     RawSuccessfulUpdateParameters successful_update = 18;
     RawFailedUpdateParameters failed_update = 19;
     RawGrowMemoryParameters grow_memory = 20;
-    RawFilesystemStorageUsageUpdateParameters filesystem_storage_usage_update = 21;
     RawCreateResourceParameters create_resource = 22;
     RawDropResourceParameters drop_resource = 23;
     RawLogParameters log = 24;
@@ -232,10 +234,6 @@ message RawGrowMemoryParameters {
   uint64 delta = 1;
 }
 
-message RawFilesystemStorageUsageUpdateParameters {
-  int64 delta = 1;
-}
-
 message RawResourceTypeId {
   string name = 1;
   string owner = 2;

golem-api-grpc/proto/golem/worker/worker_error.proto

@@ -5,6 +5,9 @@ package golem.worker;
 import "golem/common/environment.proto";
 
 message AgentError {
+  reserved 11, 12;
+  reserved "node_out_of_filesystem_storage", "agent_exceeded_filesystem_storage_limit";
+
   oneof error {
     StackOverflow stack_overflow = 1;
     OutOfMemory out_of_memory = 2;
@@ -16,8 +19,6 @@ message AgentError {
     TransientError transient_error = 8;
     PermanentError permanent_error = 9;
     ExceededTableLimit exceeded_table_limit = 10;
-    NodeOutOfFilesystemStorage node_out_of_filesystem_storage = 11;
-    AgentExceededFilesystemStorageLimit agent_exceeded_filesystem_storage_limit = 12;
     ExceededHttpCallLimit exceeded_http_call_limit = 13;
     ExceededRpcCallLimit exceeded_rpc_call_limit = 14;
     AgentTerminatedByQuota agent_terminated_by_quota = 15;
@@ -64,10 +65,6 @@ message ExceededHttpCallLimit {}
 
 message ExceededRpcCallLimit {}
 
-message NodeOutOfFilesystemStorage {}
-
-message AgentExceededFilesystemStorageLimit {}
-
 message AgentTerminatedByQuota {
   golem.common.EnvironmentId environment_id = 1;
   string resource_name = 2;

golem-common/src/base_model/oplog/mod.rs

@@ -386,18 +386,6 @@ oplog_entry! {
             delta: u64
         }
     },
-    /// Updated storage usage by a signed delta (positive = write, negative = delete/shrink)
-    FilesystemStorageUsageUpdate {
-        hint: true
-        wit_raw_type: "filesystem-storage-usage-update-parameters"
-        wit_public_type: "filesystem-storage-usage-update-parameters"
-        raw {
-            delta: i64
-        }
-        public {
-            delta: i64
-        }
-    },
     /// Created a resource instance
     CreateResource {
         hint: true

golem-common/src/model/mod.rs

@@ -700,7 +700,6 @@ pub struct AgentStatusRecord {
     pub component_revision: ComponentRevision,
     pub component_size: u64,
     pub total_linear_memory_size: u64,
-    pub current_filesystem_storage_usage: u64,
     pub owned_resources: HashMap<AgentResourceId, AgentResourceDescription>,
     pub oplog_idx: OplogIndex,
     pub active_plugins: HashSet<EnvironmentPluginGrantId>,
@@ -746,7 +745,6 @@ impl Default for AgentStatusRecord {
             component_revision: ComponentRevision::INITIAL,
             component_size: 0,
             total_linear_memory_size: 0,
-            current_filesystem_storage_usage: 0,
             owned_resources: HashMap::new(),
             oplog_idx: OplogIndex::default(),
             active_plugins: HashSet::new(),

golem-common/src/model/oplog/matcher.rs

@@ -322,10 +322,6 @@ impl PublicOplogEntry {
                 Self::string_match("growmemory", &[], query_path, query)
                     || Self::string_match("grow-memory", &[], query_path, query)
             }
-            PublicOplogEntry::FilesystemStorageUsageUpdate(_params) => {
-                Self::string_match("filesystemstorageusageupdate", &[], query_path, query)
-                    || Self::string_match("filesystem-storage-usage-update", &[], query_path, query)
-            }
             PublicOplogEntry::CreateResource(_params) => {
                 Self::string_match("createresource", &[], query_path, query)
                     || Self::string_match("create-resource", &[], query_path, query)