Mirror/seaweedfs - seaweedfs - Gitea: Git with a cup of tea

Author	SHA1	Message	Date
Chris Lu	b565a0cc86	Adds volume.merge command with deduplication and disk-based backend (#8441 ) * Enhance volume.merge command with deduplication and disk-based backend * Fix copyVolume function call with correct argument order and missing bool parameter * Revert "Fix copyVolume function call with correct argument order and missing bool parameter" This reverts commit `7b4a190643`. * Fix critical issues: per-replica writable tracking, tail goroutine cancellation via done channel, and debug logging for allocation failures * Optimize memory usage with watermark approach for duplicate detection * Fix critical issues: swap copyVolume arguments, increase idle timeout, remove file double-close, use glog for logging * Replace temporary file with in-memory buffer for needle blob serialization * test(volume.merge): Add comprehensive unit and integration tests Add 7 unit tests covering: - Ordering by timestamp - Cross-stream duplicate deduplication - Empty stream handling - Complex multi-stream deduplication - Single stream passthrough - Large needle ID support - LastModified fallback when timestamp unavailable Add 2 integration validation tests: - TestMergeWorkflowValidation: Documents 9-stage merge workflow - TestMergeEdgeCaseHandling: Validates 10 edge case handling All tests passing (9/9) * fix(volume.merge): Use time window for deduplication to handle clock skew The same needle ID can have different timestamps on different servers due to clock skew and replication lag. Needles with the same ID within a 5-second time window are now treated as duplicates (same write with timestamp variance). Key changes: - Add mergeDeduplicationWindowNs constant (5 seconds) - Replace exact timestamp matching with time window comparison - Use windowInitialized flag to properly detect window transitions - Add TestMergeNeedleStreamsTimeWindowDeduplication test This ensures that replicated writes with slight timestamp differences are properly deduplicated during merge, while separate updates to the same file ID (outside the window) are preserved. All tests passing (10/10) * test: Add volume.merge integration tests with 5 comprehensive test cases * test: integration tests for volume.merge command * Fix integration tests: use TripleVolumeCluster for volume.merge testing - Created new TripleVolumeCluster framework (cluster_triple.go) with 3 volume servers - Rebuilt weed binary with volume.merge command compiled in - Updated all 5 integration tests to use TripleVolumeCluster instead of DualVolumeCluster - Tests now properly allocate volumes on 2 servers and let merge allocate on 3rd - All 5 integration tests now pass: - TestVolumeMergeBasic - TestVolumeMergeReadonly - TestVolumeMergeRestore - TestVolumeMergeTailNeedles - TestVolumeMergeDivergentReplicas * Refactor test framework: use parameterized server count instead of hardcoded - Renamed TripleVolumeCluster to MultiVolumeCluster with serverCount parameter - Replaced hardcoded volumePort0/1/2 with slices for flexible server count - Updated StartTripleVolumeCluster as backward-compatible wrapper calling StartMultiVolumeCluster(t, profile, 3) - Made directory creation, port allocation, and server startup loop-based - Updated accessor methods (VolumeAdminAddress, VolumeGRPCAddress, etc.) to support any server count - All 5 integration tests continue to pass with new parameterized cluster framework - Enables future testing with 2, 4, 5+ volume servers by calling StartMultiVolumeCluster directly * Consolidate cluster frameworks: StartDualVolumeCluster now uses MultiVolumeCluster - Made DualVolumeCluster a type alias for MultiVolumeCluster - Updated StartDualVolumeCluster to call StartMultiVolumeCluster(t, profile, 2) - Removed duplicate code from cluster_dual.go (now just 17 lines) - All existing tests using StartDualVolumeCluster continue to work without changes - Backward compatible: existing code continues to use the old function signatures - Added wrapper functions in cluster_multi.go for StartTripleVolumeCluster - Enables unified cluster management across all test suites * Address PR review comments: improve error handling and clean up code - Replace parse error swallow with proper error return - Log cleanup and restoration errors instead of silently discarding them - Remove unused offset field from memoryBackendFile struct - Fix WriteAt buffer truncation bug to preserve trailing bytes - All unit tests passing (10/10) - Code compiles successfully * Fix PR review findings: test improvements and code quality - Add timeout to runWeedShell to prevent hanging - Add server 1 readonly status verification in tests - Assert merge fails when replicas writable (not just log output) - Replace sleep with polling for writable restoration check - Fix WriteAt stale data snapshot bug in memoryBackendFile - Fix startVolume error logging to show current server log - Fix volumePubPorts double assignment in port allocation - Rename test to reflect behavior: DoesNotDeduplicateAcrossWindows - Fix misleading dedup window comment Unit tests: 10/10 passing Binary: Compiles successfully * Fix test assumption: merge command marks volumes readonly automatically TestVolumeMergeReadonly was expecting merge to fail on writable volumes, but the merge command is designed to mark volumes readonly as part of its operation. Fixed test to verify merge succeeds on writable volumes and properly restores writable state afterward. Removed redundant Test 2 code that duplicated the new behavior. * fmt * Fix deduplication logic to correctly handle same-stream vs cross-stream duplicates The dedup map previously used only NeedleId as key, causing same-stream overwrites to be incorrectly skipped as duplicates. Changed to track which stream first processed each needle ID in the current window: - Cross-stream duplicates (same ID from different streams, within window) are skipped - Same-stream duplicates (overwrites from same stream) are kept - Map now stores: needleId -> streamIndex of first occurrence in window Added TestMergeNeedleStreamsSameStreamDuplicates to verify same-stream overwrites are preserved while cross-stream duplicates are skipped. All unit tests passing (11/11) Binary compiles successfully	1 month ago
Chris Lu	da4edb5fe6	Fix live volume move tail timestamp (#8440 ) * Improve move tail timestamp * Add move tail timestamp integration test * Simulate traffic during move	1 month ago
Chris Lu	f7c27cc81f	go fmt	1 month ago
Chris Lu	beeb375a88	Add volume server integration test suite and CI workflow (#8322 ) * docs(volume_server): add integration test development plan * test(volume_server): add integration harness and profile matrix * test(volume_server/http): add admin and options integration coverage * test(volume_server/grpc): add state and status integration coverage * test(volume_server): auto-build weed binary and harden cluster startup * test(volume_server/http): add upload read range head delete coverage * test(volume_server/grpc): expand admin lifecycle and state coverage * docs(volume_server): update progress tracker for implemented tests * test(volume_server/http): cover if-none-match and invalid-range branches * test(volume_server/grpc): add batch delete integration coverage * docs(volume_server): log latest HTTP and gRPC test coverage * ci(volume_server): run volume server integration tests in github actions * test(volume_server/grpc): add needle status configure ping and leave coverage * docs(volume_server): record additional grpc coverage progress * test(volume_server/grpc): add vacuum integration coverage * docs(volume_server): record vacuum test coverage progress * test(volume_server/grpc): add read and write needle blob error-path coverage * docs(volume_server): record data rw grpc coverage progress * test(volume_server/http): add jwt auth integration coverage * test(volume_server/grpc): add sync copy and stream error-path coverage * docs(volume_server): record jwt and sync/copy test coverage * test(volume_server/grpc): add scrub and query integration coverage * test(volume_server/grpc): add volume tail sender and receiver coverage * docs(volume_server): record scrub query and tail test progress * test(volume_server/grpc): add readonly writable and collection lifecycle coverage * test(volume_server/http): add public-port cors and method parity coverage * test(volume_server/grpc): add blob meta and read-all success path coverage * test(volume_server/grpc): expand scrub and query variation coverage * test(volume_server/grpc): add tiering and remote fetch error-path coverage * test(volume_server/http): add unchanged write and delete edge-case coverage * test(volume_server/grpc): add ping unknown and unreachable target coverage * test(volume_server/grpc): add volume delete only-empty variation coverage * test(volume_server/http): add jwt fid-mismatch auth coverage * test(volume_server/grpc): add scrub ec auto-select empty coverage * test(volume_server/grpc): stabilize ping timestamp assertion * docs(volume_server): update integration coverage progress log * test(volume_server/grpc): add tier remote backend and config variation coverage * docs(volume_server): record tier remote variation progress * test(volume_server/grpc): add incremental copy and receive-file protocol coverage * test(volume_server/http): add read path shape and if-modified-since coverage * test(volume_server/grpc): add copy-file compaction and receive-file success coverage * test(volume_server/http): add passthrough headers and static asset coverage * test(volume_server/grpc): add ping filer unreachable coverage * docs(volume_server): record copy receive and http variant progress * test(volume_server/grpc): add erasure coding maintenance and missing-path coverage * docs(volume_server): record initial erasure coding rpc coverage * test(volume_server/http): add multi-range multipart response coverage * docs(volume_server): record multi-range http coverage progress * test(volume_server/grpc): add query empty-stripe no-match coverage * docs(volume_server): record query no-match stream behavior coverage * test(volume_server/http): add upload throttling timeout and replicate bypass coverage * docs(volume_server): record upload throttling coverage progress * test(volume_server/http): add download throttling timeout coverage * docs(volume_server): record download throttling coverage progress * test(volume_server/http): add jwt wrong-cookie fid mismatch coverage * docs(volume_server): record jwt wrong-cookie mismatch coverage * test(volume_server/http): add jwt expired-token rejection coverage * docs(volume_server): record jwt expired-token coverage * test(volume_server/http): add jwt query and cookie transport coverage * docs(volume_server): record jwt token transport coverage * test(volume_server/http): add jwt token-source precedence coverage * docs(volume_server): record jwt token-source precedence coverage * test(volume_server/http): add jwt header-over-cookie precedence coverage * docs(volume_server): record jwt header cookie precedence coverage * test(volume_server/http): add jwt query-over-cookie precedence coverage * docs(volume_server): record jwt query cookie precedence coverage * test(volume_server/grpc): add setstate version mismatch and nil-state coverage * docs(volume_server): record setstate validation coverage * test(volume_server/grpc): add readonly persist-true lifecycle coverage * docs(volume_server): record readonly persist variation coverage * test(volume_server/http): add options origin cors header coverage * docs(volume_server): record options origin cors coverage * test(volume_server/http): add trace unsupported-method parity coverage * docs(volume_server): record trace method parity coverage * test(volume_server/grpc): add batch delete cookie-check variation coverage * docs(volume_server): record batch delete cookie-check coverage * test(volume_server/grpc): add admin lifecycle missing and maintenance variants * docs(volume_server): record admin lifecycle edge-case coverage * test(volume_server/grpc): add mixed batch delete status matrix coverage * docs(volume_server): record mixed batch delete matrix coverage * test(volume_server/http): add jwt-profile ui access gating coverage * docs(volume_server): record jwt ui-gating http coverage * test(volume_server/http): add propfind unsupported-method parity coverage * docs(volume_server): record propfind method parity coverage * test(volume_server/grpc): add volume configure success and rollback-path coverage * docs(volume_server): record volume configure branch coverage * test(volume_server/grpc): add volume needle status missing-path coverage * docs(volume_server): record volume needle status error-path coverage * test(volume_server/http): add readDeleted query behavior coverage * docs(volume_server): record readDeleted http behavior coverage * test(volume_server/http): add delete ts override parity coverage * docs(volume_server): record delete ts parity coverage * test(volume_server/grpc): add invalid blob/meta offset coverage * docs(volume_server): record invalid blob/meta offset coverage * test(volume_server/grpc): add read-all mixed volume abort coverage * docs(volume_server): record read-all mixed-volume abort coverage * test(volume_server/http): assert head response body parity * docs(volume_server): record head body parity assertion * test(volume_server/grpc): assert status state and memory payload completeness * docs(volume_server): record volume server status payload coverage * test(volume_server/grpc): add batch delete chunk-manifest rejection coverage * docs(volume_server): record batch delete chunk-manifest coverage * test(volume_server/grpc): add query cookie-mismatch eof parity coverage * docs(volume_server): record query cookie-mismatch parity coverage * test(volume_server/grpc): add ping master success target coverage * docs(volume_server): record ping master success coverage * test(volume_server/http): add head if-none-match conditional parity * docs(volume_server): record head if-none-match parity coverage * test(volume_server/http): add head if-modified-since parity coverage * docs(volume_server): record head if-modified-since parity coverage * test(volume_server/http): add connect unsupported-method parity coverage * docs(volume_server): record connect method parity coverage * test(volume_server/http): assert options allow-headers cors parity * docs(volume_server): record options allow-headers coverage * test(volume_server/framework): add dual volume cluster integration harness * test(volume_server/http): add missing-local read mode proxy redirect local coverage * docs(volume_server): record read mode missing-local matrix coverage * test(volume_server/http): add download over-limit replica proxy fallback coverage * docs(volume_server): record download replica fallback coverage * test(volume_server/http): add missing-local readDeleted proxy redirect parity coverage * docs(volume_server): record missing-local readDeleted mode coverage * test(volume_server/framework): add single-volume cluster with filer harness * test(volume_server/grpc): add ping filer success target coverage * docs(volume_server): record ping filer success coverage * test(volume_server/http): add proxied-loop guard download timeout coverage * docs(volume_server): record proxied-loop download coverage * test(volume_server/http): add disabled upload and download limit coverage * docs(volume_server): record disabled throttling path coverage * test(volume_server/grpc): add idempotent volume server leave coverage * docs(volume_server): record leave idempotence coverage * test(volume_server/http): add redirect collection query preservation coverage * docs(volume_server): record redirect collection query coverage * test(volume_server/http): assert admin server headers on status and health * docs(volume_server): record admin server header coverage * test(volume_server/http): assert healthz request-id echo parity * docs(volume_server): record healthz request-id parity coverage * test(volume_server/http): add over-limit invalid-vid download branch coverage * docs(volume_server): record over-limit invalid-vid branch coverage * test(volume_server/http): add public-port static asset coverage * docs(volume_server): record public static endpoint coverage * test(volume_server/http): add public head method parity coverage * docs(volume_server): record public head parity coverage * test(volume_server/http): add throttling wait-then-proceed path coverage * docs(volume_server): record throttling wait-then-proceed coverage * test(volume_server/http): add read cookie-mismatch not-found coverage * docs(volume_server): record read cookie-mismatch coverage * test(volume_server/http): add throttling timeout-recovery coverage * docs(volume_server): record throttling timeout-recovery coverage * test(volume_server/grpc): add ec generate mount info unmount lifecycle coverage * docs(volume_server): record ec positive lifecycle coverage * test(volume_server/grpc): add ec shard read and blob delete lifecycle coverage * docs(volume_server): record ec shard read/blob delete lifecycle coverage * test(volume_server/grpc): add ec rebuild and to-volume error branch coverage * docs(volume_server): record ec rebuild and to-volume branch coverage * test(volume_server/grpc): add ec shards-to-volume success roundtrip coverage * docs(volume_server): record ec shards-to-volume success coverage * test(volume_server/grpc): add ec receive and copy-file missing-source coverage * docs(volume_server): record ec receive and copy-file coverage * test(volume_server/grpc): add ec last-shard delete cleanup coverage * docs(volume_server): record ec last-shard delete cleanup coverage * test(volume_server/grpc): add volume copy success path coverage * docs(volume_server): record volume copy success coverage * test(volume_server/grpc): add volume copy overwrite-destination coverage * docs(volume_server): record volume copy overwrite coverage * test(volume_server/http): add write error-path variant coverage * docs(volume_server): record http write error-path coverage * test(volume_server/http): add conditional header precedence coverage * docs(volume_server): record conditional header precedence coverage * test(volume_server/http): add oversized combined range guard coverage * docs(volume_server): record oversized range guard coverage * test(volume_server/http): add image resize and crop read coverage * docs(volume_server): record image transform coverage * test(volume_server/http): add chunk-manifest expansion and bypass coverage * docs(volume_server): record chunk-manifest read coverage * test(volume_server/http): add compressed read encoding matrix coverage * docs(volume_server): record compressed read matrix coverage * test(volume_server/grpc): add tail receiver source replication coverage * docs(volume_server): record tail receiver replication coverage * test(volume_server/grpc): add tail sender large-needle chunking coverage * docs(volume_server): record tail sender chunking coverage * test(volume_server/grpc): add ec-backed volume needle status coverage * docs(volume_server): record ec-backed needle status coverage * test(volume_server/grpc): add ec shard copy from peer success coverage * docs(volume_server): record ec shard copy success coverage * test(volume_server/http): add chunk-manifest delete child cleanup coverage * docs(volume_server): record chunk-manifest delete cleanup coverage * test(volume_server/http): add chunk-manifest delete failure-path coverage * docs(volume_server): record chunk-manifest delete failure coverage * test(volume_server/grpc): add ec shard copy source-unavailable coverage * docs(volume_server): record ec shard copy source-unavailable coverage * parallel	1 month ago