You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Tree:
f096b067fd
add-admin-and-worker-to-helm-charts
add-ec-vacuum
add-foundation-db
add_fasthttp_client
add_remote_storage
adding-message-queue-integration-tests
avoid_releasing_temp_file_on_write
changing-to-zap
collect-public-metrics
create-table-snapshot-api-design
data_query_pushdown
dependabot/github_actions/helm/kind-action-1.13.0
dependabot/go_modules/cloud.google.com/go/storage-1.57.1
dependabot/go_modules/github.com/aws/aws-sdk-go-v2-1.39.5
dependabot/go_modules/github.com/aws/aws-sdk-go-v2/credentials-1.18.20
dependabot/go_modules/github.com/aws/aws-sdk-go-v2/service/s3-1.89.1
dependabot/go_modules/go.mongodb.org/mongo-driver-1.17.6
dependabot/maven/other/java/client/com.google.protobuf-protobuf-java-3.25.5
dependabot/maven/other/java/examples/org.apache.hadoop-hadoop-common-3.4.0
detect-and-plan-ec-tasks
do-not-retry-if-error-is-NotFound
enhance-erasure-coding
fasthttp
filer1_maintenance_branch
fix-GetObjectLockConfigurationHandler
fix-versioning-listing-only
ftp
gh-pages
improve-fuse-mount
improve-fuse-mount2
logrus
master
message_send
mount2
mq-subscribe
mq2
original_weed_mount
pr-7412
random_access_file
refactor-needle-read-operations
refactor-volume-write
remote_overlay
revert-5134-patch-1
revert-5819-patch-1
revert-6434-bugfix-missing-s3-audit
s3-select
sub
tcp_read
test-reverting-lock-table
test_udp
testing
testing-sdx-generation
tikv
track-mount-e2e
volume_buffered_writes
worker-execute-ec-tasks
0.72
0.72.release
0.73
0.74
0.75
0.76
0.77
0.90
0.91
0.92
0.93
0.94
0.95
0.96
0.97
0.98
0.99
1.00
1.01
1.02
1.03
1.04
1.05
1.06
1.07
1.08
1.09
1.10
1.11
1.12
1.14
1.15
1.16
1.17
1.18
1.19
1.20
1.21
1.22
1.23
1.24
1.25
1.26
1.27
1.28
1.29
1.30
1.31
1.32
1.33
1.34
1.35
1.36
1.37
1.38
1.40
1.41
1.42
1.43
1.44
1.45
1.46
1.47
1.48
1.49
1.50
1.51
1.52
1.53
1.54
1.55
1.56
1.57
1.58
1.59
1.60
1.61
1.61RC
1.62
1.63
1.64
1.65
1.66
1.67
1.68
1.69
1.70
1.71
1.72
1.73
1.74
1.75
1.76
1.77
1.78
1.79
1.80
1.81
1.82
1.83
1.84
1.85
1.86
1.87
1.88
1.90
1.91
1.92
1.93
1.94
1.95
1.96
1.97
1.98
1.99
1;70
2.00
2.01
2.02
2.03
2.04
2.05
2.06
2.07
2.08
2.09
2.10
2.11
2.12
2.13
2.14
2.15
2.16
2.17
2.18
2.19
2.20
2.21
2.22
2.23
2.24
2.25
2.26
2.27
2.28
2.29
2.30
2.31
2.32
2.33
2.34
2.35
2.36
2.37
2.38
2.39
2.40
2.41
2.42
2.43
2.47
2.48
2.49
2.50
2.51
2.52
2.53
2.54
2.55
2.56
2.57
2.58
2.59
2.60
2.61
2.62
2.63
2.64
2.65
2.66
2.67
2.68
2.69
2.70
2.71
2.72
2.73
2.74
2.75
2.76
2.77
2.78
2.79
2.80
2.81
2.82
2.83
2.84
2.85
2.86
2.87
2.88
2.89
2.90
2.91
2.92
2.93
2.94
2.95
2.96
2.97
2.98
2.99
3.00
3.01
3.02
3.03
3.04
3.05
3.06
3.07
3.08
3.09
3.10
3.11
3.12
3.13
3.14
3.15
3.16
3.18
3.19
3.20
3.21
3.22
3.23
3.24
3.25
3.26
3.27
3.28
3.29
3.30
3.31
3.32
3.33
3.34
3.35
3.36
3.37
3.38
3.39
3.40
3.41
3.42
3.43
3.44
3.45
3.46
3.47
3.48
3.50
3.51
3.52
3.53
3.54
3.55
3.56
3.57
3.58
3.59
3.60
3.61
3.62
3.63
3.64
3.65
3.66
3.67
3.68
3.69
3.71
3.72
3.73
3.74
3.75
3.76
3.77
3.78
3.79
3.80
3.81
3.82
3.83
3.84
3.85
3.86
3.87
3.88
3.89
3.90
3.91
3.92
3.93
3.94
3.95
3.96
3.97
3.98
3.99
4.00
dev
helm-3.65.1
v0.69
v0.70beta
v3.33
${ noResults }
Chris Lu
d745e6e41d
* fallback to check master
* clean up
* parsing
* refactor
* handle parse error
* return error
* avoid dup lookup
* use batch key
* dedup lookup logic
* address comments
* errors.Join(lookupErrors...)
* add a comment
* Fix: Critical data race in MasterClient vidMap
Fixes a critical data race where resetVidMap() was writing to the vidMap
pointer while other methods were reading it concurrently without synchronization.
Changes:
- Removed embedded *vidMap from MasterClient
- Added vidMapLock (sync.RWMutex) to protect vidMap pointer access
- Created safe accessor methods (GetLocations, GetDataCenter, etc.)
- Updated all direct vidMap accesses to use thread-safe methods
- Updated resetVidMap() to acquire write lock during pointer swap
The vidMap already has internal locking for its operations, but this fix
protects the vidMap pointer itself from concurrent read/write races.
Verified with: go test -race ./weed/wdclient/...
Impact:
- Prevents potential panics from concurrent pointer access
- No performance impact - uses RWMutex for read-heavy workloads
- Maintains backward compatibility through wrapper methods
* fmt
* Fix: Critical data race in MasterClient vidMap
Fixes a critical data race where resetVidMap() was writing to the vidMap
pointer while other methods were reading it concurrently without synchronization.
Changes:
- Removed embedded *vidMap from MasterClient struct
- Added vidMapLock (sync.RWMutex) to protect vidMap pointer access
- Created minimal public accessor methods for external packages:
* GetLocations, GetLocationsClone, GetVidLocations
* LookupFileId, LookupVolumeServerUrl
* GetDataCenter
- Internal code directly locks and accesses vidMap (no extra indirection)
- Updated resetVidMap() to acquire write lock during pointer swap
- Updated shell/commands.go to use GetDataCenter() method
Design Philosophy:
- vidMap already has internal locking for its map operations
- This fix specifically protects the vidMap *pointer* from concurrent access
- Public methods for external callers, direct locking for internal use
- Minimizes wrapper overhead while maintaining thread safety
Verified with: go test -race ./weed/wdclient/... (passes)
Impact:
- Prevents potential panics/crashes from data races
- Minimal performance impact (RWMutex for read-heavy workload)
- Maintains full backward compatibility
* fix more concurrent access
* reduce lock scope
* Optimize vidMap locking for better concurrency
Improved locking strategy based on the understanding that:
- vidMapLock protects the vidMap pointer from concurrent swaps
- vidMap has internal locks that protect its data structures
Changes:
1. Read operations: Grab pointer with RLock, release immediately, then operate
- Reduces lock hold time
- Allows resetVidMap to proceed sooner
- Methods: GetLocations, GetLocationsClone, GetVidLocations,
LookupVolumeServerUrl, GetDataCenter
2. Write operations: Changed from Lock() to RLock()
- RLock prevents pointer swap during operation
- Allows concurrent readers and other writers (serialized by vidMap's lock)
- Methods: addLocation, deleteLocation, addEcLocation, deleteEcLocation
Benefits:
- Significantly reduced lock contention
- Better concurrent performance under load
- Still prevents all race conditions
Verified with: go test -race ./weed/wdclient/... (passes)
* Further reduce lock contention in LookupVolumeIdsWithFallback
Optimized two loops that were holding RLock for extended periods:
Before:
- Held RLock during entire loop iteration
- Included string parsing and cache lookups
- Could block resetVidMap for significant time with large batches
After:
- Grab vidMap pointer with brief RLock
- Release lock immediately
- Perform all loop operations on local pointer
Impact:
- First loop: Cache check on initial volumeIds
- Second loop: Double-check after singleflight wait
Benefits:
- Minimal lock hold time (just pointer copy)
- resetVidMap no longer blocked by long loops
- Better concurrent performance with large volume ID lists
- Still thread-safe (vidMap methods have internal locks)
Verified with: go test -race ./weed/wdclient/... (passes)
* Add clarifying comments to vidMap helper functions
Added inline documentation to each helper function (addLocation, deleteLocation,
addEcLocation, deleteEcLocation) explaining the two-level locking strategy:
- RLock on vidMapLock prevents resetVidMap from swapping the pointer
- vidMap has internal locks that protect the actual map mutations
- This design provides optimal concurrency
The comments make it clear why RLock (not Lock) is correct and intentional,
preventing future confusion about the locking strategy.
* Improve encapsulation: Add shallowClone() method to vidMap
Added a shallowClone() method to vidMap to improve encapsulation and prevent
MasterClient from directly accessing vidMap's internal fields.
Changes:
1. Added vidMap.shallowClone() in vid_map.go
- Encapsulates the shallow copy logic within vidMap
- Makes vidMap responsible for its own state representation
- Documented that caller is responsible for thread safety
2. Simplified resetVidMap() in masterclient.go
- Uses tail := mc.vidMap.shallowClone() instead of manual field access
- Cleaner, more maintainable code
- Better adherence to encapsulation principles
Benefits:
- Improved code organization and maintainability
- vidMap internals are now properly encapsulated
- Easier to modify vidMap structure in the future
- More self-documenting code
Verified with: go test -race ./weed/wdclient/... (passes)
* Optimize locking: Reduce lock acquisitions and use helper methods
Two optimizations to further reduce lock contention and improve code consistency:
1. LookupFileIdWithFallback: Eliminated redundant lock acquisition
- Before: Two separate locks to get vidMap and dataCenter
- After: Single lock gets both values together
- Benefit: 50% reduction in lock/unlock overhead for this hot path
2. KeepConnected: Use GetDataCenter() helper for consistency
- Before: Manual lock/unlock to access DataCenter field
- After: Use existing GetDataCenter() helper method
- Benefit: Better encapsulation and code consistency
Impact:
- Reduced lock contention in high-traffic lookup path
- More consistent use of accessor methods throughout codebase
- Cleaner, more maintainable code
Verified with: go test -race ./weed/wdclient/... (passes)
* Refactor: Extract common locking patterns into helper methods
Eliminated code duplication by introducing two helper methods that encapsulate
the common locking patterns used throughout MasterClient:
1. getStableVidMap() - For read operations
- Acquires lock, gets pointer, releases immediately
- Returns stable snapshot for thread-safe reads
- Used by: GetLocations, GetLocationsClone, GetVidLocations,
LookupFileId, LookupVolumeServerUrl, GetDataCenter
2. withCurrentVidMap(f func(vm *vidMap)) - For write operations
- Holds RLock during callback execution
- Prevents pointer swap while allowing concurrent operations
- Used by: addLocation, deleteLocation, addEcLocation, deleteEcLocation
Benefits:
- Reduced code duplication (eliminated 48 lines of repetitive locking code)
- Centralized locking logic makes it easier to understand and maintain
- Self-documenting pattern through named helper methods
- Easier to modify locking strategy in the future (single point of change)
- Improved readability - accessor methods are now one-liners
Code size reduction: ~40% fewer lines for accessor/helper methods
Verified with: go test -race ./weed/wdclient/... (passes)
* consistent
* Fix cache pointer race condition with atomic.Pointer
Use atomic.Pointer for vidMap cache field to prevent data races
during cache trimming in resetVidMap. This addresses the race condition
where concurrent GetLocations calls could read the cache pointer while
resetVidMap is modifying it during cache chain trimming.
Changes:
- Changed cache field from *vidMap to atomic.Pointer[vidMap]
- Updated all cache access to use Load() and Store() atomic operations
- Updated shallowClone, GetLocations, deleteLocation, deleteEcLocation
- Updated resetVidMap to use atomic operations for cache trimming
* Merge: Resolve conflict in deleteEcLocation - keep atomic.Pointer and fix bug
Resolved merge conflict by combining:
1. Atomic pointer access pattern (from HEAD): cache.Load()
2. Correct method call (from fix): deleteEcLocation (not deleteLocation)
Resolution:
- Before (HEAD): cachedMap.deleteLocation() - WRONG, reintroduced bug
- Before (fix): vc.cache.deleteEcLocation() - RIGHT method, old pattern
- After (merged): cachedMap.deleteEcLocation() - RIGHT method, new pattern
This preserves both improvements:
✓ Thread-safe atomic.Pointer access pattern
✓ Correct recursive call to deleteEcLocation
Verified with: go test -race ./weed/wdclient/... (passes)
* Update vid_map.go
* remove shallow clone
* simplify
|
4 days ago | |
|---|---|---|
| .. | ||
| exclusive_locks | ExclusiveLocker only create one renew goroutine (#6269) | 12 months ago |
| net2 | convert error fromating to %w everywhere (#6995) | 4 months ago |
| resource_pool | convert error fromating to %w everywhere (#6995) | 4 months ago |
| masterclient.go | Fix masterclient vidmap race condition (#7412) | 4 days ago |
| vid_map.go | Fix masterclient vidmap race condition (#7412) | 4 days ago |
| vid_map_test.go | Implement SRV lookups for filer (#4767) | 2 years ago |