issue: #44358
Implement complete snapshot management system including creation,
deletion, listing, description, and restoration capabilities across all
system components.
Key features:
- Create snapshots for entire collections
- Drop snapshots by name with proper cleanup
- List snapshots with collection filtering
- Describe snapshot details and metadata
Components added/modified:
- Client SDK with full snapshot API support and options
- DataCoord snapshot service with metadata management
- Proxy layer with task-based snapshot operations
- Protocol buffer definitions for snapshot RPCs
- Comprehensive unit tests with mockey framework
- Integration tests for end-to-end validation
Technical implementation:
- Snapshot metadata storage in etcd with proper indexing
- File-based snapshot data persistence in object storage
- Garbage collection integration for snapshot cleanup
- Error handling and validation across all operations
- Thread-safe operations with proper locking mechanisms
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
- Core invariant/assumption: snapshots are immutable point‑in‑time
captures identified by (collection, snapshot name/ID); etcd snapshot
metadata is authoritative for lifecycle (PENDING → COMMITTED → DELETING)
and per‑segment manifests live in object storage (Avro / StorageV2). GC
and restore logic must see snapshotRefIndex loaded
(snapshotMeta.IsRefIndexLoaded) before reclaiming or relying on
segment/index files.
- New capability added: full end‑to‑end snapshot subsystem — client SDK
APIs (Create/Drop/List/Describe/Restore + restore job queries),
DataCoord SnapshotWriter/Reader (Avro + StorageV2 manifests),
snapshotMeta in meta, SnapshotManager orchestration
(create/drop/describe/list/restore), copy‑segment restore
tasks/inspector/checker, proxy & RPC surface, GC integration, and
docs/tests — enabling point‑in‑time collection snapshots persisted to
object storage and restorations orchestrated across components.
- Logic removed/simplified and why: duplicated recursive
compaction/delta‑log traversal and ad‑hoc lookup code were consolidated
behind two focused APIs/owners (Handler.GetDeltaLogFromCompactTo for
delta traversal and SnapshotManager/SnapshotReader for snapshot I/O).
MixCoord/coordinator broker paths were converted to thin RPC proxies.
This eliminates multiple implementations of the same traversal/lookup,
reducing divergence and simplifying responsibility boundaries.
- Why this does NOT introduce data loss or regressions: snapshot
create/drop use explicit two‑phase semantics (PENDING → COMMIT/DELETING)
with SnapshotWriter writing manifests and metadata before commit; GC
uses snapshotRefIndex guards and
IsRefIndexLoaded/GetSnapshotBySegment/GetSnapshotByIndex checks to avoid
removing referenced files; restore flow pre‑allocates job IDs, validates
resources (partitions/indexes), performs rollback on failure
(rollbackRestoreSnapshot), and converts/updates segment/index metadata
only after successful copy tasks. Extensive unit and integration tests
exercise pending/deleting/GC/restore/error paths to ensure idempotence
and protection against premature deletion.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Signed-off-by: Wei Liu <wei.liu@zilliz.com>
related: #45993
Add nullable vector support in import utility layer
Key changes:
ImportV2 util:
- Add nullable vector types (FloatVector, Float16Vector, BFloat16Vector,
BinaryVector, SparseFloatVector, Int8Vector) to
AppendNullableDefaultFieldsData()
- Add tests for nullable vector field data appending
CSV/JSON/Numpy readers:
- Add nullPercent parameter to test data generation for better null
coverage
- Mark vector fields as nullable in test schemas
- Add test cases for nullable vector field parsing
- Refactor tests to use loop-based approach with 0%, 50%, 100% null
percentages
Parquet field reader:
- Add ReadNullableBinaryData() for nullable
BinaryVector/Float16Vector/BFloat16Vector
- Add ReadNullableFloatVectorData() for nullable FloatVector
- Add ReadNullableSparseFloatVectorData() for nullable SparseFloatVector
- Add ReadNullableInt8VectorData() for nullable Int8Vector
- Add ReadNullableStructData() for generic nullable struct data
- Update Next() to use nullable read methods when field is nullable
- Add null data validation for non-nullable fields
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
- Core invariant: import must preserve per-row alignment and validity
for every field — nullable vector fields are expected to be encoded with
per-row validity masks and all readers/writers must emit arrays aligned
to original input rows (null entries represented explicitly).
- New feature & scope: adds end-to-end nullable-vector support in the
import utility layer — AppendNullableDefaultFieldsData in
internal/datanode/importv2/util.go now appends nil placeholders for
nullable vectors (FloatVector, Float16Vector, BFloat16Vector,
BinaryVector, SparseFloatVector, Int8Vector); parquet reader
(internal/util/importutilv2/parquet/field_reader.go) adds
ReadNullableBinaryData, ReadNullableFloatVectorData,
ReadNullableSparseFloatVectorData, ReadNullableInt8VectorData,
ReadNullableStructData and routes nullable branches to these helpers;
CSV/JSON/Numpy readers and test utilities updated to generate and
validate 0/50/100% null scenarios and mark vector fields as nullable in
test schemas.
- Logic removed / simplified: eliminates ad-hoc "parameter-invalid"
rejections for nullable vectors inside FieldReader.Next by centralizing
nullable handling into ReadNullable* helpers and shared validators
(getArrayDataNullable,
checkNullableVectorAlignWithDim/checkNullableVectorAligned), simplifying
control flow and removing scattered special-case checks.
- No data loss / no regression (concrete code paths): nulls are
preserved end-to-end — AppendNullableDefaultFieldsData explicitly
inserts nil entries per null row (datanode import append path);
ReadNullable*Data helpers return both data and []bool validity masks so
callers in field_reader.go and downstream readers receive exact per-row
validity; testutil.BuildSparseVectorData was extended to accept
validData so sparse vectors are materialized only for valid rows while
null rows are represented as missing. These concrete paths ensure null
rows are represented rather than dropped, preventing data loss or
behavioral regression.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Signed-off-by: marcelo-cjl <marcelo.chen@zilliz.com>
Related to #44956
This change propagates the useLoonFFI configuration through the import
pipeline to enable LOON FFI usage during data import operations.
Key changes:
- Add use_loon_ffi field to ImportRequest protobuf message
- Add manifest_path field to ImportSegmentInfo for tracking manifest
- Initialize manifest path when creating segments (both import and
growing)
- Pass useLoonFFI flag through NewSyncTask in import tasks
- Simplify pack_writer_v2 by removing GetManifestInfo method and relying
on pre-initialized manifest path from segment creation
- Update segment meta with manifest path after import completion
This allows the import workflow to use the LOON FFI based packed writer
when the common.useLoonFFI configuration is enabled.
---------
Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
issue: #43745
Add timestamp filtering capability to L0Reader to match the
functionality available in the regular Reader. This enhancement allows
filtering delete records based on timestamp range during L0 import
operations.
Changes include:
- Add tsStart and tsEnd fields to l0Reader struct for timestamp
filtering
- Modify NewL0Reader function signature to accept tsStart and tsEnd
parameters
- Implement timestamp filtering logic in Read method to skip records
outside the specified range
- Update L0ImportTask and L0PreImportTask to parse timestamp parameters
from request options and pass them to NewL0Reader
- Add comprehensive test case TestL0Reader_ReadWithTsFilter to verify ts
filtering functionality using mockey framework
Signed-off-by: Wei Liu <wei.liu@zilliz.com>
1. Use blocking memory allocation to wait until memory becomes available
2. Perform memory allocation at the file level instead of per task
3. Limit Parquet file reader batch size to prevent excessive memory
consumption
4. Limit import buffer size from 20% to 10% of total memory
issue: https://github.com/milvus-io/milvus/issues/43387,
https://github.com/milvus-io/milvus/issues/43131
---------
Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
Ref https://github.com/milvus-io/milvus/issues/42148https://github.com/milvus-io/milvus/pull/42406 impls the segcore part of
storage for handling with VectorArray.
This PR:
1. impls the go part of storage for VectorArray
2. impls the collection creation with StructArrayField and VectorArray
3. insert and retrieve data from the collection.
---------
Signed-off-by: SpadeA <tangchenjie1210@gmail.com>
Signed-off-by: SpadeA-Tang <tangchenjie1210@gmail.com>
Signed-off-by: SpadeA-Tang <u6748471@anu.edu.au>
issue: #43072, #43289
- manage the schema version at recovery storage.
- update the schema when creating collection or alter schema.
- get schema at write buffer based on version.
- recover the schema when upgrading from 2.5.
---------
Signed-off-by: chyezh <chyezh@outlook.com>
1. Modify the binlog reader to stop reading a fixed 4096 rows and
instead use the calculated bufferSize to avoid generating small binlogs.
2. Use a fixed bufferSize (32MB) for the Parquet reader to prevent OOM.
issue: https://github.com/milvus-io/milvus/issues/43387
---------
Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
- Introduce dynamic buffer sizing to avoid generating small binlogs
during import
- Refactor import slot calculation based on CPU and memory constraints
- Implement dynamic pool sizing for sync manager and import tasks
according to CPU core count
issue: https://github.com/milvus-io/milvus/issues/43131
---------
Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
1. Optimize the import process: skip subsequent steps and mark the task
as complete if the number of imported rows is 0.
2. Improve import integration tests:
a. Add a test to verify that autoIDs are not duplicated
b. Add a test for the corner case where all data is deleted
c. Shorten test execution time
3. Enhance import logging:
a. Print imported segment information upon completion
b. Include file name in failure logs
issue: https://github.com/milvus-io/milvus/issues/42488,
https://github.com/milvus-io/milvus/issues/42518
Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
Increase insert buffer size from 16MB to 64MB, while keeping delete
buffer size at 16MB.
issue: https://github.com/milvus-io/milvus/issues/42518
Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
Remove the unlimited logID mechanism and switch to redundantly
allocating a large number of IDs.
issue: https://github.com/milvus-io/milvus/issues/42518
Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
issue: #41976
- make drop partition message as a broadcast message.
- add gc when drop partition message is acked.
- add a call back to handle the broadcast message when ack.
- the ack operation of broadcast message will retry until success.
Signed-off-by: chyezh <chyezh@outlook.com>
1. Add global scheduler for datacoord.
2. Define and implement new CreateTask, QueryTask, DropTask interfaces.
3. Refine Import, Compaction, Stats, Index task.
issue: https://github.com/milvus-io/milvus/issues/41123
Co-authored-by: Cai Zhang <cai.zhang@zilliz.com>
Close the chunk manager's reader after the import completes to prevent
goroutine leaks.
issues: https://github.com/milvus-io/milvus/issues/41868
---------
Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
When autoID is enabled, the preimport task estimates row distribution by
evenly dividing the total row count (numRows) across all vchannels:
`estimatedCount = numRows / vchannelNum`.
However, the actual import task hashes real auto-generated IDs to
determine
the target vchannel. This mismatch can lead to inaccurate row
distribution estimation
in such corner cases:
- Importing 1 row into 2 vchannels:
• Preimport: 1 / 2 = 0 → both v0 and v1 are estimated to have 0 rows
• Import: real autoID (e.g., 457975852966809057) hashes to v1
→ actual result: v0 = 0, v1 = 1
To resolve such corner case, we now allocate at least one segment for
each vchannel
when autoID is enabled, ensuring all vchannels are prepared to receive
data even
if no rows are estimated for them.
issue: https://github.com/milvus-io/milvus/issues/41759
---------
Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
Related to #35415
In rolling upgrade, legacy proxy may dispatch load request wit empty
load field list. The upgraded querycoord may report error by mistake
that load field list is changed.
This PR:
- Auto field empty load field list with all user field ids
- Refine the error messag when load field list updates
- Refine load job unit test with service cases
Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
Related to #35303#30404
This PR change return type of `DeleteCodec.Deserialize` from
`storage.DeleteData` to `DeltaData`, which
reduces the memory usage of interface header.
Also refine `storage.DeltaData` methods to make it easier to usage.
Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
issue: #36621
1. Add API to access task runtime metrics, including:
- build index task
- compaction task
- import task
- balance (including load/release of segments/channels and some leader
tasks on querycoord)
- sync task
2. Add a debug model to the webpage by using debug=true or debug=false
in the URL query parameters to enable or disable debug mode.
Signed-off-by: jaime <yun.zhang@zilliz.com>
