Before these changes the request tracing was added for each route
registered using the routing.RouteRegister, see code. This had the
consequence that middleware executed earlier/later in the request
pipeline was not part of the request tracing middleware life-cycle
which measures the duration of requests among other things.
In the logger middleware we do extract the current distributed trace
identifier, if available, and set that on request info/error log messages.
With these changes we can extract the current distributed trace identifier,
if available, and set that on the contextual HTTP request logger
(models.ReqContext.Logger) which would improve the possibility to correlate
all HTTP request log messages with traces.
In addition, the request tracing middleware is now executed first and last in
the request pipeline and should therefore result in more accurate timing
measurements (request duration).
Co-authored-by: Arve Knudsen <arve.knudsen@gmail.com>
* Move db package WIP
* Implement OSS access control
* Register OSS access control
* Fix linter error in tests
* Fix linter error in evaluator
* Simplify OSS tests
* Optimize builtin roles
* Chore: add comments to the exported functions
* Remove init from ossaccesscontrol package (moved to ext)
* Add access control as a dependency for http server
* Modify middleware to receive fallback function
* Middleware: refactor fallback function call
* Move unused models to enterprise
* Simplify AccessControl type
* Chore: use bool IsDisabled() method instead of CanBeDisabled interface
This pull request migrates testdata to coreplugin streaming capabilities,
this is mostly a working concept of streaming plugins at the moment,
the work will continue in the following pull requests.
* Fix LoginService.UpsertUser user creation
* Fix API AdminCreateUser user creation
* Add missing underscore import
* Fix API CompleteInvite user creation
* Fix API SignUpStep2 user creation
* Introduce PluginRequestValidator abstraction with a NoOp implementation
* Update PluginRequestValidator abstraction to use the dsURL instead
* Inject PluginRequestValidator into the HTTPServer and validate requests going through data source proxy
* Inject PluginRequestValidator into the BackendPluginManager and validate requests going through it
* Validate requests going through QueryMetrics & QueryMetricsV2
* Validate BackendPluginManager health requests
* Fix backend plugins manager tests
* Validate requests going through alerting service
* Fix tests
* fix tests
* goimports
Co-authored-by: Leonard Gram <leo@xlson.com>
* middleware: Move context handler to own service
Signed-off-by: Arve Knudsen <arve.knudsen@gmail.com>
Co-authored-by: Emil Tullsted <sakjur@users.noreply.github.com>
Co-authored-by: Will Browne <wbrowne@users.noreply.github.com>
* Alerting NG: prototype v2 (WIP)
* Separate eval package
* Modify eval alert definition endpoint
* Disable migration if ngalert is not enabled
* Remove premature test
* Fix lint issues
* Delete obsolete struct
* Apply suggestions from code review
* Update pkg/services/ngalert/ngalert.go
Co-authored-by: Kyle Brandt <kyle@grafana.com>
* Add API endpoint for listing alert definitions
* Introduce index for alert_definition table
* make ds object for expression to avoid panic
* wrap error
* Update pkg/services/ngalert/eval/eval.go
* Swith to backend.DataQuery
* Export TransformWrapper callback
* Fix lint issues
* Update pkg/services/ngalert/ngalert.go
Co-authored-by: Kyle Brandt <kyle@grafana.com>
* Validate alert definitions before storing them
* Introduce AlertQuery
* Add test
* Add QueryType in AlertQuery
* Accept only float64 (seconds) durations
* Apply suggestions from code review
* Get rid of bus
* Do not export symbols
* Fix failing test
* Fix failure due to service initialization order
Introduce MediumHigh service priority and assign it to backendplugin
service
* Fix test
* Apply suggestions from code review
* Fix renamed reference
Co-authored-by: Kyle Brandt <kyle@grafana.com>
* Backend: Adds route for well-known change password URL
* Include 'dashboard/new' in backend routes
* Move index route handler registration out of "not logged in views" section
Co-authored-by: Arve Knudsen <arve.knudsen@gmail.com>
* intial frontend resolution/redirection logic
* backend scaffolding
* enough of the frontend to actually test end to end
* bugfixes
* add tests
* cleanup
* explore too hard for now
* fix build
* Docs: add docs
* FE test
* redirect directly from backend
* validate incoming uids
* add last_seen_at
* format documentation
* more documentation feedback
* very shaky migration of get route to middleware
* persist unix timestamps
* add id, orgId to table
* fixes for orgId scoping
* whoops forgot the middleware
* only redirect to absolute URLs under the AppUrl domain
* move lookup route to /goto/:uid, stop manually setting 404 response code
* renaming things according to PR feedback
* tricky deletion
* sneaky readd
* fix test
* more BE renaming
* FE updates -- no more @ts-ignore hacking :) and accounting for subpath
* Simplify code
Signed-off-by: Arve Knudsen <arve.knudsen@gmail.com>
* Short URLs: Drop usage of bus
Signed-off-by: Arve Knudsen <arve.knudsen@gmail.com>
* ShortURLService: Make injectable
Signed-off-by: Arve Knudsen <arve.knudsen@gmail.com>
* Rename file
Signed-off-by: Arve Knudsen <arve.knudsen@gmail.com>
* Add handling of url parsing and creating of full shortURL to backend
* Update test, remove unused imports
* Update pkg/api/short_urls.go
Co-authored-by: Arve Knudsen <arve.knudsen@gmail.com>
* Add correct import
* Pass context to short url service
* Remove not needed error log
* Rename dto and field to denote URL rather than path
* Update api docs based on feedback/suggestion
* Rename files to singular
* Revert to send relative path to backend
* Fixes after review
* Return dto when creating short URL that includes the full url
Use full url to provide shorten URL to the user
* Fix after review
* Fix relative url path when creating new short url
Co-authored-by: Arve Knudsen <arve.knudsen@gmail.com>
Co-authored-by: Ivana <ivana.huckova@gmail.com>
Co-authored-by: Ivana Huckova <30407135+ivanahuckova@users.noreply.github.com>
Co-authored-by: Marcus Efraimsson <marcus.efraimsson@gmail.com>
* New feature toggle for enabling alerting NG
* Initial commit
* Modify evaluate alert API request
* Check for unique labels in alert execution result dataframes
* Remove print statement
* Additional minor fixes/comments
* Fix lint issues
* Add API endpoint for evaluating panel queries
* Push missing renaming
* add refId for condition to API
* add refId for condition to API
* switch dashboard based eval to get method
* add from/to params to dashboard based eval
* add from/to params to eval endpoint
Co-authored-by: kyle <kyle@grafana.com>
kuberentes (and Im sure other orchastrators does as well) support two
kind of checks. readiness checks and liveness checks. Grafanas current
`/api/health` endpoint requires database access which might not
always be required for the instance to be considered active.
* Settings: Add setting for hiding version number for anonymous users
Fixes#12925
* Hide version string from footer when unavailable
* Settings: Test frontend settings with hide version for anonymous users
* Settings: Add hide version variable to frontend settings
* Make AnonymousHideVersion non-global
Signed-off-by: Arve Knudsen <arve.knudsen@gmail.com>
* Settings: Improve test neighbor friendliness, reset state before and after
* Settings: Use T.Cleanup
Co-authored-by: Arve Knudsen <arve.knudsen@gmail.com>
By storing render key in remote cache it will enable
image renderer to use public facing url or load
balancer url to render images and thereby remove
the requirement of image renderer having to use the
url of the originating Grafana instance when running
HA setup (multiple Grafana instances).
Fixes#17704
Ref grafana/grafana-image-renderer#91
Put the cipher suites with Forward Secrecy at or nearer the top, keeping any TLS v1.3 suites at the top, following best practice guides for the ordering of the rest. There is no change to the selection of suites only reordering.
* Licensing: supplies a service to handle licensing information
* Licensing: uses the license service further
Uses the license service instead of settings.isEnterprise:
- external team members
- saml
- usage stats
* Licensing: fixes broken tests due to new Licensing service dependency
* Licensing: fixes linting errors
* Licensing: exposes license expiry information to the frontend
* LDAP: use only one struct
* Use only models.ExternalUserInfo
* Add additional helper method :/
* Move all the helpers to one module
* LDAP: refactoring
* Rename some of the public methods and change their behaviour
* Remove outdated methods
* Simplify logic
* More tests
There is no and never were tests for settings.go, added tests for helper
methods (cover is now about 100% for them). Added tests for the main
LDAP logic, but there is some stuff to add. Dial() is not tested and not
decoupled. It might be a challenge to do it properly
* Restructure tests:
* they wouldn't depend on external modules
* more consistent naming
* logical division
* More guards for erroneous paths
* Login: make login service an explicit dependency
* LDAP: remove no longer needed test helper fns
* LDAP: remove useless import
* LDAP: Use new interface in multildap module
* LDAP: corrections for the groups of multiple users
* In case there is several users their groups weren't detected correctly
* Simplify helpers module
Adds a new [server] setting `serve_from_sub_path`. By enabling
this setting and using a subpath in `root_url` setting, e.g.
`root_url = http://localhost:3000/grafana`, Grafana will be accessible
on `http://localhost:3000/grafana`. By default it is set to `false`
for compatibility reasons.
Closes#16623
Currently all API requests set Cache-control: no-cache to avoid browsers
caching sensitive data. This fixes so that all responses returned from
backend not are cached using http headers. The exception is the data proxy
where we don't add these http headers in case datasource backend needs
to control whether data can be cached or not.
Fixes#16845
* Add file path to metadata and show it in dialogs
* Make path relative to config directory
* Fix tests
* Add test for the relative path
* Refactor to use path relative to provisioner path
* Change return types
* Rename attribute
* Small fixes from review
* Add api to reaload provisioning
* Refactor and simplify the polling code
* Add test for the provisioning service
* Fix provider initialization and move some code to file reader
* Simplify the code and move initialization
* Remove unused code
* Update comment
* Add comment
* Change error messages
* Add DashboardProvisionerFactory type
* Update imports
* Use new assert lib
* Use mutext for synchronizing the reloading
* Fix typo
Co-Authored-By: aocenas <mr.ocenas@gmail.com>
* Add docs about the new api
* enable ee build on pr/master
* step1: of including group sync
* disable commit pinning for now
* fixes broken build
* enable team to ldap group sync
* avoid returning error for missing external handler
* services: allow routes to be added before http server start
* services: allows services to add their own migrations
* moves db migrations to ee code base
* build using master branch in ee
* disable enterprise build in .bra.toml
[skip ci]
* removes team sync extensions
* removes commented line
* rendering: headless chrome progress
* renderer: minor change
* grpc: version hell
* updated grpc libs
* wip: minor progess
* rendering: new image rendering plugin is starting to work
* feat: now phantomjs works as well and updated alerting to use new rendering service
* refactor: renamed renderer package and service to rendering to make renderer name less confusing (rendering is internal service that handles the renderer plugin now)
* rendering: now render key is passed and render auth is working in plugin mode
* removed unneeded lines from gitignore
* rendering: now plugin mode supports waiting for all panels to complete rendering
* fix: LastSeenAt fix for render calls, was not set which causes a lot of updates to Last Seen at during rendering, this should fix sqlite db locked issues in seen in previous releases
* change: changed render tz url parameter to use proper timezone name as chrome does not handle UTC offset TZ values
* fix: another update to tz param generation
* renderer: added http mode to renderer service, new ini setting [rendering] server_url
* wip: start on refactoring settings
* settings: progress on settings refactor
* refactor: progress on settings refactoring
* fix: fixed failing test
* settings: moved smtp settings from global to instance
* extensions: import and build
* bus: use predefined error
* enterprise: build script for enterprise packages
* poc: auto registering services and dependency injection
(cherry picked from commit b5b1ef875f905473af41e49f8071cb9028edc845)
* poc: backend services registry progress
(cherry picked from commit 97be69725881241bfbf1e7adf0e66801d6b0af3d)
* poc: minor update
(cherry picked from commit 03d7a6888b81403f458b94305792e075568f0794)
* ioc: introduce manuel ioc
* enterprise: adds setting for enterprise
* build: test and build specific ee commit
* cleanup: test testing code
* removes example hello service
This fixes:
pkg/api/http_server.go:142:89: should use make(map[string]func(*http.Server, *tls.Conn, http.Handler)) instead (S1019)
pkg/services/alerting/scheduler.go:18:30: should use make(map[int64]*Job) instead (S1019)
pkg/services/alerting/scheduler.go:26:31: should use make(map[int64]*Job) instead (S1019)
This changes forks the mysql part of the Macaron session middleware.
In the forked mysql file:
- takes in a config setting for SetConnMaxLifetime (this solves wait_timeout
problem if it is set to a shorter interval than wait_timeout)
- removes the panic when an error is returned in the Exist function.
- retries the exist query once
- retries the GC query once