fix: critical clear() data persistence regression (v5.10.4)

Workshop team reported that brain.clear() doesn't fully delete persistent storage.
After calling clear() and creating a new Brainy instance, all data was restored
from storage. This is a CRITICAL data integrity bug.

Root causes (3 bugs fixed):
1. **FileSystemStorage deleting wrong directory**: Data stored in branches/main/entities/
   but clear() was only deleting old pre-v5.4.0 structure (nouns/, verbs/, metadata/)
2. **COW reinitialization after clear()**: Setting cowEnabled=false on old instance
   doesn't affect new instances. Fixed with persistent marker file.
3. **Metadata index cache not cleared**: find() with type filters returned stale data
   after clear(). Fixed by recreating MetadataIndexManager.

Changes:
- FileSystemStorage: Clear branches/ directory (where data actually lives)
- All storage adapters: Add checkClearMarker()/createClearMarker() methods
- BaseStorage: Check for cow-disabled marker before initializing COW
- Brainy: Recreate metadataIndex after clear() to flush cached data
- Tests: Comprehensive regression suite (8 tests) to prevent recurrence

Fixes Workshop bug report: /media/dpsifr/storage/home/Projects/workshop/BRAINY_V5_10_2_CLEAR_BUG.md

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
David Snelling 2025-11-17 10:44:35 -08:00
parent 53420f6c9a
commit aba15638dc
11 changed files with 613 additions and 17 deletions

View file

@ -2196,7 +2196,7 @@ export class Brainy<T = any> implements BrainyInterface<T> {
return this.augmentationRegistry.execute('clear', {}, async () => {
// Clear storage
await this.storage.clear()
// Reset index
if ('clear' in this.index && typeof this.index.clear === 'function') {
await this.index.clear()
@ -2204,10 +2204,15 @@ export class Brainy<T = any> implements BrainyInterface<T> {
// Recreate index if no clear method
this.index = this.setupIndex()
}
// v5.10.4: Recreate metadata index to clear cached data
// Bug: Metadata index cache was not being cleared, causing find() with type filters to return stale data
this.metadataIndex = new MetadataIndexManager(this.storage)
await this.metadataIndex.init()
// Reset dimensions
this.dimensions = undefined
// Clear any cached sub-APIs
this._neural = undefined
this._nlp = undefined

View file

@ -1153,6 +1153,12 @@ export class AzureBlobStorage extends BaseStorage {
this.commitLog = undefined
this.cowEnabled = false
// v5.10.4: Create persistent marker blob (CRITICAL FIX)
// Bug: cowEnabled = false only affects current instance, not future instances
// Fix: Create marker blob that persists across instance restarts
// When new instance calls initializeCOW(), it checks for this marker
await this.createClearMarker()
// Clear caches
this.nounCacheManager.clear()
this.verbCacheManager.clear()
@ -1204,6 +1210,47 @@ export class AzureBlobStorage extends BaseStorage {
}
}
/**
* Check if COW has been explicitly disabled via clear()
* v5.10.4: Fixes bug where clear() doesn't persist across instance restarts
* @returns true if marker blob exists, false otherwise
* @protected
*/
protected async checkClearMarker(): Promise<boolean> {
await this.ensureInitialized()
try {
const markerPath = `${this.systemPrefix}cow-disabled`
const blockBlobClient = this.containerClient!.getBlockBlobClient(markerPath)
const exists = await blockBlobClient.exists()
return exists
} catch (error) {
this.logger.warn('AzureBlobStorage.checkClearMarker: Error checking marker', error)
return false
}
}
/**
* Create marker indicating COW has been explicitly disabled
* v5.10.4: Called by clear() to prevent COW reinitialization on new instances
* @protected
*/
protected async createClearMarker(): Promise<void> {
await this.ensureInitialized()
try {
const markerPath = `${this.systemPrefix}cow-disabled`
const blockBlobClient = this.containerClient!.getBlockBlobClient(markerPath)
// Create empty marker blob
await blockBlobClient.upload(Buffer.from(''), 0, {
blobHTTPHeaders: { blobContentType: 'text/plain' }
})
} catch (error) {
this.logger.error('AzureBlobStorage.createClearMarker: Failed to create marker blob', error)
// Don't throw - marker creation failure shouldn't break clear()
}
}
/**
* Save statistics data to storage
*/

View file

@ -1001,20 +1001,13 @@ export class FileSystemStorage extends BaseStorage {
}
}
// Remove all files in the nouns directory
await removeDirectoryContents(this.nounsDir)
// Remove all files in the verbs directory
await removeDirectoryContents(this.verbsDir)
// Remove all files in the metadata directory
await removeDirectoryContents(this.metadataDir)
// Remove all files in the noun metadata directory
await removeDirectoryContents(this.nounMetadataDir)
// Remove all files in the verb metadata directory
await removeDirectoryContents(this.verbMetadataDir)
// v5.10.4: Clear the entire branches/ directory (branch-based storage)
// Bug fix: Data is stored in branches/main/entities/, not just entities/
// The branch-based structure was introduced for COW support
const branchesDir = path.join(this.rootDir, 'branches')
if (await this.directoryExists(branchesDir)) {
await removeDirectoryContents(branchesDir)
}
// Remove all files in both system directories
await removeDirectoryContents(this.systemDir)
@ -1037,6 +1030,12 @@ export class FileSystemStorage extends BaseStorage {
this.blobStorage = undefined
this.commitLog = undefined
this.cowEnabled = false
// v5.10.4: Create persistent marker file (CRITICAL FIX)
// Bug: cowEnabled = false only affects current instance, not future instances
// Fix: Create marker file that persists across instance restarts
// When new instance calls initializeCOW(), it checks for this marker
await this.createClearMarker()
}
// Clear the statistics cache
@ -1075,6 +1074,49 @@ export class FileSystemStorage extends BaseStorage {
return result
}
/**
* Check if COW has been explicitly disabled via clear()
* v5.10.4: Fixes bug where clear() doesn't persist across instance restarts
* @returns true if marker file exists, false otherwise
* @protected
*/
protected async checkClearMarker(): Promise<boolean> {
// Check if fs module is available
if (!fs || !fs.promises) {
return false
}
try {
const markerPath = path.join(this.systemDir, 'cow-disabled')
await fs.promises.access(markerPath, fs.constants.F_OK)
return true // Marker exists
} catch (error) {
return false // Marker doesn't exist (ENOENT) or can't be accessed
}
}
/**
* Create marker indicating COW has been explicitly disabled
* v5.10.4: Called by clear() to prevent COW reinitialization on new instances
* @protected
*/
protected async createClearMarker(): Promise<void> {
// Check if fs module is available
if (!fs || !fs.promises) {
console.warn('FileSystemStorage.createClearMarker: fs module not available, skipping marker creation')
return
}
try {
const markerPath = path.join(this.systemDir, 'cow-disabled')
// Create empty marker file
await fs.promises.writeFile(markerPath, '', 'utf8')
} catch (error) {
console.error('FileSystemStorage.createClearMarker: Failed to create marker file', error)
// Don't throw - marker creation failure shouldn't break clear()
}
}
/**
* Get information about storage usage and capacity
*/

View file

@ -1015,6 +1015,12 @@ export class GcsStorage extends BaseStorage {
this.commitLog = undefined
this.cowEnabled = false
// v5.10.4: Create persistent marker object (CRITICAL FIX)
// Bug: cowEnabled = false only affects current instance, not future instances
// Fix: Create marker object that persists across instance restarts
// When new instance calls initializeCOW(), it checks for this marker
await this.createClearMarker()
// Clear caches
this.nounCacheManager.clear()
this.verbCacheManager.clear()
@ -1069,6 +1075,45 @@ export class GcsStorage extends BaseStorage {
}
}
/**
* Check if COW has been explicitly disabled via clear()
* v5.10.4: Fixes bug where clear() doesn't persist across instance restarts
* @returns true if marker object exists, false otherwise
* @protected
*/
protected async checkClearMarker(): Promise<boolean> {
await this.ensureInitialized()
try {
const markerPath = `${this.systemPrefix}cow-disabled`
const file = this.bucket!.file(markerPath)
const [exists] = await file.exists()
return exists
} catch (error) {
this.logger.warn('GCSStorage.checkClearMarker: Error checking marker', error)
return false
}
}
/**
* Create marker indicating COW has been explicitly disabled
* v5.10.4: Called by clear() to prevent COW reinitialization on new instances
* @protected
*/
protected async createClearMarker(): Promise<void> {
await this.ensureInitialized()
try {
const markerPath = `${this.systemPrefix}cow-disabled`
const file = this.bucket!.file(markerPath)
// Create empty marker object
await file.save('', { contentType: 'text/plain' })
} catch (error) {
this.logger.error('GCSStorage.createClearMarker: Failed to create marker object', error)
// Don't throw - marker creation failure shouldn't break clear()
}
}
/**
* Save statistics data to storage
*/

View file

@ -310,6 +310,25 @@ export class HistoricalStorageAdapter extends BaseStorage {
}
}
/**
* Check if COW has been explicitly disabled via clear()
* v5.10.4: No-op for HistoricalStorageAdapter (read-only, doesn't manage COW)
* @returns Always false (read-only adapter doesn't manage COW state)
* @protected
*/
protected async checkClearMarker(): Promise<boolean> {
return false // Read-only adapter - COW state managed by underlying storage
}
/**
* Create marker indicating COW has been explicitly disabled
* v5.10.4: No-op for HistoricalStorageAdapter (read-only)
* @protected
*/
protected async createClearMarker(): Promise<void> {
// No-op: HistoricalStorageAdapter is read-only, doesn't create markers
}
// ============= Override Write Methods (Read-Only) =============
/**

View file

@ -196,6 +196,26 @@ export class MemoryStorage extends BaseStorage {
}
}
/**
* Check if COW has been explicitly disabled via clear()
* v5.10.4: No-op for MemoryStorage (doesn't persist)
* @returns Always false (marker doesn't persist in memory)
* @protected
*/
protected async checkClearMarker(): Promise<boolean> {
return false // MemoryStorage doesn't persist - marker doesn't survive restart
}
/**
* Create marker indicating COW has been explicitly disabled
* v5.10.4: No-op for MemoryStorage (doesn't persist)
* @protected
*/
protected async createClearMarker(): Promise<void> {
// No-op: MemoryStorage doesn't persist, so marker is not needed
// clear() in memory already resets all state, no marker survives restart
}
/**
* Save statistics data to storage
* @param statistics The statistics data to save

View file

@ -495,6 +495,12 @@ export class OPFSStorage extends BaseStorage {
this.blobStorage = undefined
this.commitLog = undefined
this.cowEnabled = false
// v5.10.4: Create persistent marker file (CRITICAL FIX)
// Bug: cowEnabled = false only affects current instance, not future instances
// Fix: Create marker file that persists across instance restarts
// When new instance calls initializeCOW(), it checks for this marker
await this.createClearMarker()
} catch (error: any) {
// Ignore if _cow directory doesn't exist (not all instances use COW)
if (error.name !== 'NotFoundError') {
@ -516,6 +522,53 @@ export class OPFSStorage extends BaseStorage {
}
}
/**
* Check if COW has been explicitly disabled via clear()
* v5.10.4: Fixes bug where clear() doesn't persist across instance restarts
* @returns true if marker file exists, false otherwise
* @protected
*/
protected async checkClearMarker(): Promise<boolean> {
await this.ensureInitialized()
try {
// Get system directory (may not exist yet)
const systemDir = await this.rootDir!.getDirectoryHandle('system', { create: false })
// Try to get the marker file
await systemDir.getFileHandle('cow-disabled', { create: false })
return true // Marker exists
} catch (error: any) {
if (error.name === 'NotFoundError') {
return false // Marker doesn't exist (or system dir doesn't exist)
}
// Other errors (permissions, etc.) - treat as marker not existing
console.warn('OPFSStorage.checkClearMarker: Error checking marker', error)
return false
}
}
/**
* Create marker indicating COW has been explicitly disabled
* v5.10.4: Called by clear() to prevent COW reinitialization on new instances
* @protected
*/
protected async createClearMarker(): Promise<void> {
await this.ensureInitialized()
try {
// Get or create system directory
const systemDir = await this.rootDir!.getDirectoryHandle('system', { create: true })
// Create empty marker file
const fileHandle = await systemDir.getFileHandle('cow-disabled', { create: true })
const writable = await fileHandle.createWritable()
await writable.write(new Uint8Array(0)) // Empty file
await writable.close()
} catch (error) {
console.error('OPFSStorage.createClearMarker: Failed to create marker file', error)
// Don't throw - marker creation failure shouldn't break clear()
}
}
// Quota monitoring configuration (v4.0.0)
private quotaWarningThreshold = 0.8 // Warn at 80% usage
private quotaCriticalThreshold = 0.95 // Critical at 95% usage

View file

@ -1051,6 +1051,12 @@ export class R2Storage extends BaseStorage {
this.commitLog = undefined
this.cowEnabled = false
// v5.10.4: Create persistent marker object (CRITICAL FIX)
// Bug: cowEnabled = false only affects current instance, not future instances
// Fix: Create marker object that persists across instance restarts
// When new instance calls initializeCOW(), it checks for this marker
await this.createClearMarker()
this.nounCacheManager.clear()
this.verbCacheManager.clear()
@ -1086,6 +1092,43 @@ export class R2Storage extends BaseStorage {
}
}
/**
* Check if COW has been explicitly disabled via clear()
* v5.10.4: Fixes bug where clear() doesn't persist across instance restarts
* @returns true if marker object exists, false otherwise
* @protected
*/
protected async checkClearMarker(): Promise<boolean> {
await this.ensureInitialized()
try {
const markerPath = `${this.systemPrefix}cow-disabled`
const data = await this.readObjectFromPath(markerPath)
return data !== null // Marker exists if we got any data
} catch (error) {
prodLog.warn('R2Storage.checkClearMarker: Error checking marker', error)
return false
}
}
/**
* Create marker indicating COW has been explicitly disabled
* v5.10.4: Called by clear() to prevent COW reinitialization on new instances
* @protected
*/
protected async createClearMarker(): Promise<void> {
await this.ensureInitialized()
try {
const markerPath = `${this.systemPrefix}cow-disabled`
// Create empty marker object
await this.writeObjectToPath(markerPath, '')
} catch (error) {
prodLog.error('R2Storage.createClearMarker: Failed to create marker object', error)
// Don't throw - marker creation failure shouldn't break clear()
}
}
// v5.4.0: Removed getNounsWithPagination override - use BaseStorage's type-first implementation

View file

@ -2137,6 +2137,12 @@ export class S3CompatibleStorage extends BaseStorage {
this.commitLog = undefined
this.cowEnabled = false
// v5.10.4: Create persistent marker object (CRITICAL FIX)
// Bug: cowEnabled = false only affects current instance, not future instances
// Fix: Create marker object that persists across instance restarts
// When new instance calls initializeCOW(), it checks for this marker
await this.createClearMarker()
// Clear the statistics cache
this.statisticsCache = null
this.statisticsModified = false
@ -2261,6 +2267,62 @@ export class S3CompatibleStorage extends BaseStorage {
}
}
/**
* Check if COW has been explicitly disabled via clear()
* v5.10.4: Fixes bug where clear() doesn't persist across instance restarts
* @returns true if marker object exists, false otherwise
* @protected
*/
protected async checkClearMarker(): Promise<boolean> {
await this.ensureInitialized()
try {
const { HeadObjectCommand } = await import('@aws-sdk/client-s3')
const markerKey = `${this.systemPrefix}cow-disabled`
await this.s3Client!.send(
new HeadObjectCommand({
Bucket: this.bucketName,
Key: markerKey
})
)
return true // Marker exists
} catch (error: any) {
if (error.name === 'NotFound' || error.$metadata?.httpStatusCode === 404) {
return false // Marker doesn't exist
}
prodLog.warn('S3CompatibleStorage.checkClearMarker: Error checking marker', error)
return false
}
}
/**
* Create marker indicating COW has been explicitly disabled
* v5.10.4: Called by clear() to prevent COW reinitialization on new instances
* @protected
*/
protected async createClearMarker(): Promise<void> {
await this.ensureInitialized()
try {
const { PutObjectCommand } = await import('@aws-sdk/client-s3')
const markerKey = `${this.systemPrefix}cow-disabled`
// Create empty marker object
await this.s3Client!.send(
new PutObjectCommand({
Bucket: this.bucketName,
Key: markerKey,
Body: Buffer.from(''),
ContentType: 'text/plain'
})
)
} catch (error) {
prodLog.error('S3CompatibleStorage.createClearMarker: Failed to create marker object', error)
// Don't throw - marker creation failure shouldn't break clear()
}
}
// Batch update timer ID
protected statisticsBatchUpdateTimerId: NodeJS.Timeout | null = null
// Flag to indicate if statistics have been modified since last save

View file

@ -300,6 +300,15 @@ export abstract class BaseStorage extends BaseStorageAdapter {
branch?: string
enableCompression?: boolean
}): Promise<void> {
// v5.10.4: Check for persistent marker file (CRITICAL FIX)
// Bug: Setting cowEnabled = false on OLD instance doesn't affect NEW instances
// Fix: Check storage for persistent marker created by clear()
// If marker exists, COW was explicitly disabled and should NOT reinitialize
const markerExists = await this.checkClearMarker()
if (markerExists) {
return // COW was disabled by clear() - don't recreate _cow/ directory
}
// v5.6.1: If COW was explicitly disabled (e.g., via clear()), don't reinitialize
// This prevents automatic recreation of COW data after clear() operations
if (this.cowEnabled === false) {
@ -1704,6 +1713,23 @@ export abstract class BaseStorage extends BaseStorageAdapter {
*/
public abstract clear(): Promise<void>
/**
* Check if COW has been explicitly disabled via clear()
* v5.10.4: Fixes bug where clear() doesn't persist across instance restarts
* Each adapter checks for a marker file/object (e.g., "_system/cow-disabled")
* @returns true if COW was disabled by clear(), false otherwise
* @protected
*/
protected abstract checkClearMarker(): Promise<boolean>
/**
* Create marker indicating COW has been explicitly disabled
* v5.10.4: Called by clear() to prevent COW reinitialization on new instances
* Each adapter creates a marker file/object (e.g., "_system/cow-disabled")
* @protected
*/
protected abstract createClearMarker(): Promise<void>
/**
* Get information about storage usage and capacity
* This method should be implemented by each specific adapter