cd5d0f9eac17a360d4f1975d6873c3f725d68f2b
11 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Paul Rastoin
|
d640b93096 |
Improve v2 and cache invalidation perfs (#15467)
# Introduction Log are debug logs of `packages/twenty-server/test/integration/metadata/suites/object-metadata/create-delete-and-create-object-metadata-v2.integration-spec.ts`run ten times in a row on clean db reset ## Next Will improve cache computation to lighter invalidation. RelationLoad `query` does not seem to work with typeorm so I'll continue the custom integration i've started in https://github.com/twentyhq/twenty/tree/optimize-cache-read-v2 ## Integration tests duration Significant test duration improvement too ### Before <img width="2632" height="1402" alt="image" src="https://github.com/user-attachments/assets/2f1f0ccf-44de-4856-bfe1-4f45a351763a" /> ### After <img width="2632" height="1402" alt="image" src="https://github.com/user-attachments/assets/4bc9e6db-3046-48b9-b903-1464053936c4" /> ## What's next - The legacy cache invalidation removal - Factorizing redis calls in only one operation ## Autogenerated performance comparison ( including mutation refactor too ) [Before](https://gist.github.com/prastoin/3c1e21fa9e3b3ce4b0716902ff4a2dd6) [After](https://gist.github.com/prastoin/7bfddd14bfded2e4991a9378970a026d) The optimized implementation shows **dramatic performance improvements** across all metrics: - 🚀 **Cache Invalidation**: 156.3ms → 76.8ms (**50.9% faster**) - 🚀 **Builder Operations**: 21.3ms → 14.2ms (**33.3% faster**) - ⚡ **Consistency**: 16.4% more predictable performance --- ## 1. Overall Performance Summary | Component | Before (avg) | After (avg) | Best (After) | Worst (After) | Improvement | |-----------|-------------|-------------|--------------|---------------|-------------| | **Total Execution Time** | 180.2ms | 110.5ms | 52.3ms | 585.9ms | **38.7% faster** ⚡⚡ | | **Cache Invalidation** | 156.3ms | 76.8ms | 47.8ms | 285.1ms | **50.9% faster** ⚡⚡⚡ | | **Transaction Execution** | 22.4ms | 22.1ms | 0.99ms | 314.2ms | Similar | | **Initial Cache Retrieval** | 0.81ms | 2.08ms | 0.21ms | 8.24ms | Similar | | **Entity Builder (total)** | 21.3ms | 14.2ms | 0.36ms | 42.5ms | **33.3% faster** ⚡ | ### Total Execution Time Distribution #### Before (Legacy Sequential) ``` Time (ms) Count Percentage Visualization < 150 18 16% ████ 150-180 32 29% ███████ 180-210 35 32% ████████ 210-250 17 15% ████ 250-350 6 5% █ > 350 2 2% ▌ ``` #### After (Optimized Parallel) ``` Time (ms) Count Percentage Visualization < 70 28 31% ████████ 70-100 31 34% █████████ 100-150 18 20% █████ 150-200 8 9% ██ 200-300 4 4% █ > 300 2 2% ▌ ``` --- ## 2. Builder Performance Breakdown ### Field Metadata Builder | Operation | Before (avg) | After (avg) | Improvement | |-----------|-------------|-------------|-------------| | Matrix computation | 3.5ms | 3.4ms | Similar | | Creation validation | 15.2ms | 2.1ms | **86% faster** ⚡⚡⚡ | | Deletion validation | 0.08ms | 0.06ms | Similar | | Update validation | 1.3ms | 0.09ms | **93% faster** ⚡⚡⚡ | | Entity processing | 18.6ms | 11.8ms | **37% faster** ⚡ | | **Total validateAndBuild** | **21.3ms** | **14.2ms** | **33% faster** ⚡ | #### Performance Distribution ``` Before: ▁▂▄█████▆▄▂▁ (wide spread, 15-28ms range) After: ▁▁▃█████▃▁▁ (tight clustering, 10-18ms range) ``` ## 4. Cache Invalidation Performance Breakdown ### Cache Invalidation Summary | Metric | Before (Legacy) | After (Optimized) | Improvement | |--------|-----------------|-------------------|-------------| | **Best Time** | 131.965ms | 47.833ms | **63.7% faster** ⚡⚡⚡ | | **10th Percentile** | 140.2ms | 51.7ms | **63.1% faster** ⚡⚡⚡ | | **25th Percentile** | 146.1ms | 54.4ms | **62.7% faster** ⚡⚡⚡ | | **Median (50th)** | 155.1ms | 63.4ms | **59.1% faster** ⚡⚡⚡ | | **Average** | 156.3ms | 76.8ms | **50.9% faster** ⚡⚡⚡ | | **75th Percentile** | 160.2ms | 90.2ms | **43.7% faster** ⚡⚡ | | **90th Percentile** | 191.7ms | 100.2ms | **47.7% faster** ⚡⚡ | | **95th Percentile** | 235.6ms | 110.3ms | **53.2% faster** ⚡⚡⚡ | | **99th Percentile** | 278.2ms | 224.9ms | **19.1% faster** ⚡ | | **Worst Time** | 383.914ms | 285.102ms | **25.7% faster** ⚡ | ### Cache Invalidation Time Distribution #### Before (Legacy Sequential) ``` Time (ms) Count Percentage Visualization 130-140 3 3% ▊ 140-150 15 14% ████ 150-160 48 44% ███████████ 160-180 31 28% ███████ 180-220 8 7% ██ 220-280 3 3% ▊ > 280 2 2% ▌ ``` #### After (Optimized Parallel + Intersection) ``` Time (ms) Count Percentage Visualization < 50 3 3% ▊ 50-60 27 25% ███████ 60-70 25 23% ██████ 70-90 25 23% ██████ 90-100 15 14% ████ 100-120 8 7% ██ 120-150 3 3% ▊ > 150 4 4% █ ``` ## 6. Performance Consistency Analysis ### Standard Deviation & Variance | Metric | Before | After | Improvement | |--------|--------|-------|-------------| | **Cache Invalidation Std Dev** | 42.1ms | 35.2ms | **16.4% more consistent** ⚡ | | **Total Execution Std Dev** | 68.3ms | 89.1ms | Slightly more variable | | **Coefficient of Variation (Cache)** | 26.9% | 45.8% | More variance | | **Outliers (> 2σ)** | 5 cases | 3 cases | **40% fewer outliers** ⚡ | --- |
||
|
Paul Rastoin
|
267af42412 |
Centralize v2 errors types in twenty-shared (#15358)
# Introduction Followup of https://github.com/twentyhq/twenty/pull/15331 ( Reducing size by concerns ) This PR centralizes v2 format error types in `twenty-shared` and consuming them in the existing v2 error format logic in `twenty-server` ## Next This https://github.com/twentyhq/twenty/pull/15360 handles the frontend v2 format error refactor ## Conclusion Related to https://github.com/twentyhq/core-team-issues/issues/1776 |
||
|
Paul Rastoin
|
7d747c9876 |
[REQUIRES_CACHE_FLUSH][GQL_VIEW_GROUP_API_BREAKING_CHANGE] ViewGroup in v2 (#15052)
# Introduction Adding view-group to core engine v2 Following https://github.com/twentyhq/twenty/pull/15010 ( same pattern ) ## What's done - Created flat-view-group - flat view group runner - flat view group builder - create view group service v2 and input transpilers - refactor the existing view group resolver to fix standard ( BREAKING_CHANGE on graphql api update especially ) REST stays the same - refactored the front to consume the mutations autogenerated close https://github.com/twentyhq/core-team-issues/issues/1665 |
||
|
Paul Rastoin
|
6188c72f74 |
Simplify and enhance v2 type devxp (#15032)
# Introduction This PR introduces a huge type refactor that will leverage dynamic intra entity optimistic flat maps update in the future and also a more granular cache invalidation enhancing performances close https://github.com/twentyhq/core-team-issues/issues/1717 close https://github.com/twentyhq/core-team-issues/issues/1716 close https://github.com/twentyhq/core-team-issues/issues/1643 ## What's done ### Comparators centralization Comparator is now done through global configuration as const for each metadata names Thanks to Note: Definition of standard is evolving, standard is now scoped to an app. Meaning that a manifest should be able to update its own standards objects but on other app standards ones ? Each synchronizable entities will have a standardOverrides ? ## Typing refactor ### `AllFlatEntityTypesByMetadataName` **Single source of truth for the complete type ecosystem**, mapping each metadata name to its entity types, flat entities, and migration actions: ```typescript export type AllFlatEntityTypesByMetadataName = { fieldMetadata: { actions: { created: CreateFieldAction; updated: UpdateFieldAction; deleted: DeleteFieldAction; }; flatEntity: FlatFieldMetadata; entity: FieldMetadataEntity; }; objectMetadata: { /* ... */ }; // ... all 10 metadata types }; ``` ### `ALL_METADATA_NAME_MANY_TO_ONE_RELATIONS` **Explicitly declares database relationships** between entities with compile-time validation: ```typescript export const ALL_METADATA_NAME_MANY_TO_ONE_RELATIONS = { viewField: { view: 'viewId', fieldMetadata: 'fieldMetadataId', }, cronTrigger: { serverlessFunction: 'serverlessFunctionId', }, // ... all relations } as const satisfies MetadataNameAndRelations; ``` ### `ALL_FLAT_ENTITY_CONFIGURATION` **Centralizes comparison and serialization logic** for each metadata type: ```typescript export const ALL_FLAT_ENTITY_CONFIGURATION = { fieldMetadata: { propertiesToCompare: ['name', 'type', 'label', 'defaultValue', /* ... */], propertiesToStringify: ['options', 'settings', 'defaultValue'], }, objectMetadata: { propertiesToCompare: ['nameSingular', 'namePlural', 'isActive', /* ... */], propertiesToStringify: [], }, // ... all metadata types } as const satisfies AllFlatEntityConfiguration; ``` ## Combined Impact These three configurations work together to create a **strongly-typed, centrally-managed metadata system**: 1. **`AllFlatEntityTypesByMetadataName`** defines *what exists* 2. **`ALL_METADATA_NAME_MANY_TO_ONE_RELATIONS`** defines *how they relate* 3. **`ALL_FLAT_ENTITY_CONFIGURATION`** defines *how to compare and serialize them* **Result:** Builders and validators become thin wrappers around type-safe, configuration-driven logic instead of containing scattered, error-prone manual implementations. ## What's next ### StandardOverrides standardization Every metadata entity can be a standard one for a workspace if it's an installed app, which means it might not expose the whole entity api to be editable through an import dynamically The standard overrides logic should not be applied to Fields and Objects but to every entities At the moment we have a logic of `EDITABLE_PROPERTIES` through the api, and also `STANDARD_OVERRIDEDABLE_PROPERTIES` This should be configuration centered like `propertiesToCompare` and `propertiesToStringify`. Scoping this PR to two last for the moment. As update dispatch to standardOverrides could be considered as a side effect prefer waiting to start the side effect refactor ### Granular Optimistic deprecation With this new grain at runtime we will be able to add a flat entity and dispatch its addition to related flat maps, so we don't have to describe an optimistic method for each flat entity operations See `addFlatEntityToFlatEntityAndRelatedEntityMapsOrThrow` Note: Still in wip and included in this PR but about to create a new one to integrate these utils and remove existing methods ### ValidateBuildAndRun dynamic args typed defintion We should restrain the devxp to send expected flat maps entity as at least from to or dependency as we now have the grain both a type lvl and runtime to do so It should not be possible in the devxp to forgot adding the views to the v2 builder when passing the view field anymore ( that would lead to permanent validation error in view field integrity checks ) ## Conclusion Thanks for reading and reviewing ! Any suggestions are more than welcomed ! ( same as for questions too ! ) |
||
|
Paul Rastoin
|
651ab184a7 |
[GQL_VIEW_FILTER_API_BREAKING_CHANGE][WHEN_RELEASED_REQUIRES_CACHE_FLUSH] ViewFilter migration to workspace migration v2 (#15010)
# Introduction Migrating `viewFilter` to v2 in order to migrate later the field update side effect on view to v2 too ## What's done - Created flat-view-filter - flat view filter runner - flat view filter builder - create view filter service v2 and input transpilers - refactor the existing view filter resolver to fix standard ( BREAKING_CHANGE on graphql api update especially ) REST stays the same - refactored the front to consume the mutations autogenerated ## New generic tools ### Compare two flat entity Introducing a new util to compare two flat entity, it's strictly typed and will be added to the generic builder in a following PR This will ease flat entity addition as won't required to create a specific abstraction for comparison Generic builder will expect specific constant: properties to compare and properties to stringify ### Transform flat entity for comparison Forked and refactor the initial existing method for flat entity business scope and type safety ## Coverage Migrated existing integration tests to fit new contract API This PR does not add strong coverage on validation exceptions Deadlines are too short close https://github.com/twentyhq/core-team-issues/issues/1666 |
||
|
Paul Rastoin
|
59fbe35a8c |
Move view in metadata-modules/ and create atomic folder + module for each view entity (#14990)
# Introduction Preparing view-filter and view-group introduction in v2 core engine Moving view from `core-modules` to `metadata-modules` ## What happened ### Created dedicated modules for each view entity: - ViewFieldModule - ViewFilterModule - ViewFilterGroupModule - ViewGroupModule - ViewSortModule ### Each module is now completely independent with its own: - Controller - Resolver - Service - Entity ### Created dedicated abstraction metadata module folder for: - flat-view-field - flat-view ### Dependencies - Eleminated circular dep on ViewModule to all others ones - Granular import not importing the whole viewModule anymore everywhere close https://github.com/twentyhq/core-team-issues/issues/1703 |
||
|
Paul Rastoin
|
d3a7241b6f |
Cache flush after database reset (#14873)
When starting the app on a fresh database reset the cache would be
filled with empty flat field metadata maps
Because the reset command hack through the repository directly in order
to create views and stuff
implemented an iso flush as the one existing initially
added it to a workspace deletion tambien
```
{
byId: {],
universalIdById: {}
}
```
|
||
|
Paul Rastoin
|
170f6d27c5 |
Extract field metadata build out of object metadata build (#14763)
# Introduction Extracting the legacy fields build and dispatch out of the object one to follow the generic flat entity build, also update caches entries for object ## Main tasks - flat field map cache - flat field builder - refactored the dispatch matrix to return flat entity maps instead of flat entity arrays - orchestrator aggregator - removing legacy code - making universal identifier aka standardId of standard field for custom object deterministically dynamic - perfs debug logs for v2 ## TODO - [x] Refactor the generic entity builder to be dependency flat maps in order to main foreign keys list in flat parent - [x] Refactor the flat object metadata to contain the array of related fields and avoid costy find object fields - [ ] Improve the create field handler to handle multiple field at once - [ ] Refactor the dispatch to embbed the comparison - [ ] Improve perf by extracting from elements out of existing - [ ] Fix the labelIdentifierId validators on object before field creation ( integ tests are in failing mode ) ## Debug logs snippet ```ts [EntityBuilder fieldMetadata] matrix computation: 0.027ms [EntityBuilder fieldMetadata] creation validation: 0.001ms [EntityBuilder fieldMetadata] deletion validation: 0.293ms [EntityBuilder fieldMetadata] update validation: 0.006ms [EntityBuilder fieldMetadata] entity processing: 0.363ms [EntityBuilder fieldMetadata] validateAndBuild: 0.455ms [EntityBuilder index] matrix computation: 0.005ms [EntityBuilder index] creation validation: 0.001ms [EntityBuilder index] deletion validation: 0.146ms [EntityBuilder index] update validation: 0.004ms [EntityBuilder index] entity processing: 0.199ms [EntityBuilder index] validateAndBuild: 0.228ms [Runner] Initial cache retrieval: 0.549ms [BaseWorkspaceMigrationRunnerActionHandlerService] delete_index executeForWorkspaceSchema: 11.665ms [BaseWorkspaceMigrationRunnerActionHandlerService] delete_index executeForMetadata: 12.864ms [BaseWorkspaceMigrationRunnerActionHandlerService] delete_field executeForWorkspaceSchema: 1.476ms [BaseWorkspaceMigrationRunnerActionHandlerService] delete_field executeForMetadata: 6.816ms [BaseWorkspaceMigrationRunnerActionHandlerService] delete_field executeForWorkspaceSchema: 0.062ms [BaseWorkspaceMigrationRunnerActionHandlerService] delete_field executeForMetadata: 0.889ms [Runner] Transaction execution: 23.434ms [Runner] Cache invalidation: 316.662ms [Runner] Total execution: 340.767ms ``` As you can see cache invalidation is way to long, we could replace the cache by the optimistic in the end |
||
|
Paul Rastoin
|
fbd3286792 |
Index v2 side effects (#14567)
# Introduction Honestly this implem is a mess, discussing a potential side effect handler with @weiko before the build and run that would handle each side effect per entity and operation Handling: - [ ] unique index is generated when a field is updated with the `isUnique` - [x] an index is generated when a relation is created - [x] search vector index creation on custom object creation - [x] renaming a field metadata or an object should re-create all related indexes which are composed by their namings - [x] delete object should remove any related indexes - [x] delete field should update related indexes ( if index ends up empty it should be removed ) - [ ] on object renaming that contains morph fields -> triggers update field -> trigger index recompute - [x] on update name renaming should recompute all related indexes ## TODO - [x] Integration testing - [ ] Refactor the index maps cache to be storing a `idsByObjectMetadataId` - [x] Refactor deterministic name to use order sorting - [x] Remove flat index from flat object ## What's next Will handle morph indexes in a new dedicated PR for the moment will stick to this Same for the cache improvement and uniqueness |
||
|
Paul Rastoin
|
371c26bc9b |
Flat entity maps cache generic service + runner dynamically retrieving invalidating update cache + view service v2 refactor (#14508)
# Introduction Migrate previous runner only iterating on `flatObjectMetadataMaps` to `allFlatEntityMaps`. Refactored the optimistic to be handled inside the actions handler ## Workspace flat map cache Introducing a new service and registry, that will dynamically retrieve and or recompute requested cache when called ## Runner refactor Runner now dynamically invalidate updated cache at the end of the transaction close https://github.com/orgs/twentyhq/projects/1/views/8?pane=issue&itemId=129136356&issue=twentyhq%7Ccore-team-issues%7C1492 close https://github.com/orgs/twentyhq/projects/1/views/8?pane=issue&itemId=129136210&issue=twentyhq%7Ccore-team-issues%7C1494 |
||
|
Weiko
|
d9a6e7e0b1 |
Add flat map cache service (#14418)
## Context
Adding a new service that provides an abstract caching system for
FlatEntityMaps.
This takes care of cache invalidation and storing local and remote cache
for the map with retrieval after a comparison with map hash between
local and remote (redis).
## Implementation
Remote (redis) keys
- Flat map data:
`engine:workspace:flat-maps:{flatMapKey}:{workspaceId}:flat-map`
- Content hash:
`engine:workspace:flat-maps:{flatMapKey}:{workspaceId}:hash`
**Local Cache Hit**: If local hash matches Redis hash, return local data
**Remote Cache Hit**: If Redis has data with different hash, update
local cache
**Remote Cache Miss**: Recompute from database, store in Remote and
locally
**Invalidation**: Remove from Remote, triggering recomputation on next
access
## Usage
```typescript
@WorkspaceFlatMapCache('view') // redis key
export class WorkspaceFlatViewMapCacheService extends WorkspaceFlatMapCacheService<FlatViewMaps> {
constructor(
@InjectCacheStorage(CacheStorageNamespace.EngineWorkspace)
cacheStorageService: CacheStorageService,
@InjectRepository(ViewEntity)
private readonly viewRepository: Repository<ViewEntity>,
) {
super(cacheStorageService);
}
// only method to implement
public async computeFlatMap(workspaceId: string): Promise<FlatViewMaps> {
const views = await this.viewRepository.find({
where: { workspaceId },
relations: ['viewFields'],
select: { viewFields: { id: true } },
});
return generateFlatViewMaps(views);
}
}
```
```typescript
// 2 public methods, getExistingOrRecomputeFlatMaps to fetch the map and invalidateCache after a mutation
await this.workspaceFlatViewMapCacheService.invalidateCache(
viewData.workspaceId,
);
const flatViewMaps =
await this.workspaceFlatViewMapCacheService.getExistingOrRecomputeFlatMaps(
workspaceId,
);
```
## Multi-Pod Synchronization
- Each pod maintains local cache for performance
- SHA256 hash of flatMap content used for version comparison
- `invalidateCache()` reset Redis data, forcing other pods to refresh
when calling getExistingOrRecomputeFlatMaps
---
<img width="1072" height="325" alt="Screenshot 2025-09-11 at 15 04 48"
src="https://github.com/user-attachments/assets/ed6ce82c-db35-4a1b-8a4e-247b694e2ddf"
/>
<img width="1030" height="328" alt="Screenshot 2025-09-11 at 15 04 40"
src="https://github.com/user-attachments/assets/43a15789-7f27-4104-a9bb-bef19908a5cd"
/>
Next step: Implement a locking mechanism by reusing existing WithLock
decorator
|