* feat(backend): job result stream optimization
* get offset locally instead of from db
* fix: agent worker result stream
* update ref
* nit
* remove foreign key on job
* fix build
* feat: add worker_group_job_stats table for job metrics aggregation
- Add new table with hour timestamp, worker group, script lang, workspace_id, job count and total duration
- Workers accumulate stats in memory and update hourly via sum aggregation
- Monitor.rs cleans up rows older than 60 days periodically
- Stats are flushed on worker shutdown to prevent data loss
Co-authored-by: windmill-internal-app[bot] <windmill-internal-app[bot]@users.noreply.github.com>
* all
* all
* ee-repo-ref
* nits
* nits
---------
Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
Co-authored-by: windmill-internal-app[bot] <windmill-internal-app[bot]@users.noreply.github.com>
Co-authored-by: Ruben Fiszel <ruben@windmill.dev>
* s3 proxy works with get (no auth yet)
* nit
* support s3:// syntax
* Support s3:// syntax and fix vite api proxy normalizing double slashes in URI
* s3 checks authed
* nit
* PUT works
* delete file works
* Derive the JWT signature from the backend
* Authorize s3 correctly (JWT signature is never sent in cleartext)
* convert object store error to wmill error for correct status code
* stash
* fix
* POST first request proxy works
* s3 put for duckdb
* factor out direct proxy code
* Fix Issue with backend proxy and wrong signature due to Host header mismatch
* Add _default_ syntax to solve URI normalization issues with signing
* restricted to user paths toggle
* user path restriction works !
* change restriction to allow
* fix
* factor out code
* better permissions UX in object storage settings
* Revert to restrict_to_user_paths
* check permissions in old s3 api
* DuckDB now uses S3 Proxy and no longer needs LFS query
* implement todo
* fix hardcoded w_id
* s3 proxy size limit
* s3_proxy is ee
* nit
* add Google Cloud Storage as option to secondary storage
* GCS secret in duckdb
* fix toolchain compile
* Remove user permissions for v0
* fix ci 2
* fix CI OSS
* fix missing feature flag
* fix unused warning
* integration test fails bc rustc 1.85.0
* ee ref
* fix ci ...
* update ee ref
* feat(backend): retrieve root_job for MiniPulledJob + fix root job for flow jobs
* nit
* nits
* set root_job only if additional information meaning if not equal to innermost or parent
* nit
* feat: add root job env var
* nits
* fix build
* nit
* fix build
* nit
* sqlx
* feat: add 60-second cache for variables and resources with PostgreSQL invalidation
- Add new var_resource_cache module with 60-second TTL
- Implement PostgreSQL NOTIFY/LISTEN for immediate cache invalidation
- Cache get_variable() for non-secret, non-encrypted reads
- Cache get_resource_value() for all read operations
- Add database triggers on variable/resource table changes
- Initialize cache system in main.rs after database connection
- Add Clone derive to ListableVariable for cache compatibility
Performance benefits:
- Avoids database queries on cache hits
- Immediate invalidation ensures data consistency
- Selective caching respects security constraints
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* refactor: move cache to API layer with allow_cache query param
- Move variable/resource cache from windmill-common to windmill-api
- Add allow_cache query parameter to variables and resources endpoints
- Follow raw script cache pattern with timestamp + value structure
- Create proper database migration for notification triggers
- Include encrypted values in cache when appropriate
- Only activate caching when explicitly requested
Co-authored-by: Ruben Fiszel <rubenfiszel@users.noreply.github.com>
* refactor: move cache to API layer with allow_cache query param
Move PostgreSQL LISTEN logic to main.rs following established pattern
- Remove custom listener initialization from cache module
- Add cache invalidation channels to centralized notification handler
- Simplified cache module to only handle cache operations
- Follow raw script cache pattern for notification handling
Co-authored-by: Ruben Fiszel <rubenfiszel@users.noreply.github.com>
---------
Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Ruben Fiszel <rubenfiszel@users.noreply.github.com>
* feat: add 60-second cache for workspace key retrieval
This implements a cache with 60-second staleness for the get_workspace_key
function to reduce database queries for workspace encryption keys. The cache
follows the same pattern as the existing CUSTOM_ENVS_CACHE but with a
shorter expiration time.
Requested by @rubenfiszel
Co-authored-by: windmill-internal-app[bot] <windmill-internal-app[bot]@users.noreply.github.com>
* Extend cache staleness from 60 to 300 seconds
* feat: add cache invalidation notifications for workspace keys
Add PostgreSQL LISTEN/NOTIFY mechanism to invalidate workspace key cache
across all servers and workers when workspace keys change.
- Add database migration with trigger function for workspace_key changes
- Add notification handler in main.rs to remove from WORKSPACE_KEY_CACHE
- Follow same pattern as existing workspace environment cache invalidation
- Ensures distributed cache consistency for workspace encryption keys
Co-authored-by: Ruben Fiszel <rubenfiszel@users.noreply.github.com>
* finish
---------
Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
Co-authored-by: windmill-internal-app[bot] <windmill-internal-app[bot]@users.noreply.github.com>
Co-authored-by: Ruben Fiszel <ruben@windmill.dev>
Co-authored-by: Ruben Fiszel <rubenfiszel@users.noreply.github.com>
* feat: ai agent steps base
* better backend and graph
* feat: anthropic, log viewer
* nit
* fix(frontend): hide tool nodes from timeline
* move ai agent actions from flow status to flow status module
* nits and workspace/hub scripts support
* tmp ref
* fix merge
* feat: display agent tools status in the graph
* fix reactivity
* fix flow status
* nit
* feat: add prometheus metric queue_running_count
Adds a new Prometheus metric queue_running_count that tracks the number
of currently running jobs per tag, similar to the existing queue_count
metric but filtered for running=true instead of running=false.
Changes:
- Added get_queue_running_counts() function in windmill-common/src/queue.rs
- Added QUEUE_RUNNING_COUNT Prometheus metric in monitor.rs
- Added /workers/queue_running_counts API endpoint
- Updated OpenAPI specification
- Added SQLx query cache entry
Requested by @rubenfiszel
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: windmill-internal-app[bot] <windmill-internal-app[bot]@users.noreply.github.com>
* Remove database insertion for queue_running_count metrics
Keep Prometheus metrics and API endpoint functionality while removing
the database INSERT statements as requested.
Co-authored-by: Ruben Fiszel <rubenfiszel@users.noreply.github.com>
* sqlx
* improve logic
---------
Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
Co-authored-by: windmill-internal-app[bot] <windmill-internal-app[bot]@users.noreply.github.com>
Co-authored-by: Ruben Fiszel <rubenfiszel@users.noreply.github.com>
Co-authored-by: Ruben Fiszel <ruben@windmill.dev>
* upgrade duckdb
* basic ducklake works
* ducklake works with custom db catalogs
* fix: pwsh skip already installed modules outside of cache (#6037)
* improve query performance of user stats
* separate ducklake_catalog db
* ducklake settings
* DucklakeSettings frontend
* Ducklake ws settings saved in backend
* fetch ducklake catalog resource
* Ducklake works with configured s3 storage
* Ducklake as asset
* ducklake asset icon
* Fix duckdb array and object args not working properly (#6254)
* Fix bug with comments in duckdb
* Avoid multiple queries when doing ATTACH ducklake
* trunc sig no longer needed now that comments are trimmed
* cache DuckdbConnectionSettingsResponse
* duplicated code
* transform_attach_ducklake contributes to duckdb_connection_settings_cache
* eliminate the need for used_storages
* nit
* cleaner management of the bigquery credentials file
* DBManagerDrawer refactor to prepare for Ducklake
* get ducklake schema
* implement delete for ducklake
* load column metadata for ducklake
* Select query works for ducklake, basic db explorer works !
* duckdb count query
* Support all db ops for ducklake
* clean migrations
* SQL repl for Ducklake
* fix broken database studio
* nit
* assert function
* Ducklake in Editor Bar
* default ducklake syntax + allow extra args
* DucklakeCatalogWizard UI
* nit + remove extra $
* modal when databases do not exist
* cannot be windmill
* Ducklake works safely with instance database
* Avoid sending instance db credentials on network
* resource leak security
* remove fetch_attach_db_conn_str
* prevent instance pg password leak
* hide asset usage count when not available
* case unsensitivity duckdb
* warnings
* disable instance catalog
* use shorthand syntax when inserting with EditorBar
* Instance ducklake catalog is now safe to use
* use safer argon2 pwd
* update package json parsers
* update package json
* better msgs
* tooltips
* disable explore button until saved
* nit
* fix warnings
* better ducklake_user password management
* nit
* Sanitize passwords from errors in ducklake
* DisplayResult broken in job result
* remove superadmin requirement to check databases_exist
* duckdb_connection_settings_v2_inner
* Ducklake works on agent worker (finally)
* ci
* #[allow(dead_code)]
* fix openapi missing response
* Separate +Database button for DuckDB in EditorBar
* Fix dropdown in ducklake settings
* Attempt to fix migration race condition in CI
* update sqlx failing for some offline queries
* avoid temp password for ducklake_user
* nits
* ducklake settings nits
* update duckdb default script
* fix sql repl resetting text on refresh
* avoid pgcrypto extension
---------
Co-authored-by: HugoCasa <hugo@casademont.ch>
Co-authored-by: Ruben Fiszel <ruben@windmill.dev>
