1
0
Fork 0
mirror of https://github.com/chrislusf/seaweedfs synced 2025-09-10 05:12:47 +02:00
seaweedfs/weed/worker/tasks/base/volume_utils.go
Chris Lu 0ecb466eda
Admin: refactoring active topology (#7073)
* refactoring

* add ec shard size

* address comments

* passing task id

There seems to be a disconnect between the pending tasks created in ActiveTopology and the TaskDetectionResult returned by this function. A taskID is generated locally and used to create pending tasks via AddPendingECShardTask, but this taskID is not stored in the TaskDetectionResult or passed along in any way.

This makes it impossible for the worker that eventually executes the task to know which pending task in ActiveTopology it corresponds to. Without the correct taskID, the worker cannot call AssignTask or CompleteTask on the master, breaking the entire task lifecycle and capacity management feature.

A potential solution is to add a TaskID field to TaskDetectionResult and worker_pb.TaskParams, ensuring the ID is propagated from detection to execution.

* 1 source multiple destinations

* task supports multi source and destination

* ec needs to clean up previous shards

* use erasure coding constants

* getPlanningCapacityUnsafe getEffectiveAvailableCapacityUnsafe  should return StorageSlotChange for calculation

* use CanAccommodate to calculate

* remove dead code

* address comments

* fix Mutex Copying in Protobuf Structs

* use constants

* fix estimatedSize

The calculation for estimatedSize only considers source.EstimatedSize and dest.StorageChange, but omits dest.EstimatedSize. The TaskDestination struct has an EstimatedSize field, which seems to be ignored here. This could lead to an incorrect estimation of the total size of data involved in tasks on a disk. The loop should probably also include estimatedSize += dest.EstimatedSize.

* at.assignTaskToDisk(task)

* refactoring

* Update weed/admin/topology/internal.go

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

* fail fast

* fix compilation

* Update weed/worker/tasks/erasure_coding/detection.go

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

* indexes for volume and shard locations

* dedup with ToVolumeSlots

* return an additional boolean to indicate success, or an error

* Update abstract_sql_store.go

* fix

* Update weed/worker/tasks/erasure_coding/detection.go

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

* Update weed/admin/topology/task_management.go

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

* faster findVolumeDisk

* Update weed/worker/tasks/erasure_coding/detection.go

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

* Update weed/admin/topology/storage_slot_test.go

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

* refactor

* simplify

* remove unused GetDiskStorageImpact function

* refactor

* add comments

* Update weed/admin/topology/storage_impact.go

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

* Update weed/admin/topology/storage_slot_test.go

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

* Update storage_impact.go

* AddPendingTask

The unified AddPendingTask function now serves as the single entry point for all task creation, successfully consolidating the previously separate functions while maintaining full functionality and improving code organization.

---------

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
2025-08-03 01:35:38 -07:00

36 lines
1.2 KiB
Go

package base
import (
"github.com/seaweedfs/seaweedfs/weed/admin/topology"
)
// FindVolumeDisk finds the disk ID where a specific volume is located on a given server.
// Returns the disk ID and a boolean indicating whether the volume was found.
// Uses O(1) indexed lookup for optimal performance on large clusters.
//
// This is a shared utility function used by multiple task detection algorithms
// (balance, vacuum, etc.) to locate volumes efficiently.
//
// Example usage:
//
// // In balance task: find source disk for a volume that needs to be moved
// sourceDisk, found := base.FindVolumeDisk(topology, volumeID, collection, sourceServer)
//
// // In vacuum task: find disk containing volume that needs cleanup
// diskID, exists := base.FindVolumeDisk(topology, volumeID, collection, serverID)
func FindVolumeDisk(activeTopology *topology.ActiveTopology, volumeID uint32, collection string, serverID string) (uint32, bool) {
if activeTopology == nil {
return 0, false
}
// Use the new O(1) indexed lookup for better performance
locations := activeTopology.GetVolumeLocations(volumeID, collection)
for _, loc := range locations {
if loc.ServerID == serverID {
return loc.DiskID, true
}
}
// Volume not found on this server
return 0, false
}