* Revert "pkgs/tsdb/[grafana-pyroscope-datasource|parca]: Fix use of request headers in responses"
This reverts commit 8bac68e906bbb6d075da45397014fb7091fbab23.
* Profiles: Stop passing request/response headers to the backend
pkgs/tsdb/[grafana-pyroscope-datasource|parca]: Fix use of request headers in responses
In the parca and the grafana-pyroscope-datasource we were wrongly using the request headers instead of the response
header when communication the results to the backend.
This PR fixes this bug.
Was reported by an user via community slack, who faced issues, with a request header of `content-length: 0` being
inserted by a intermediate proxy.
* Bump tempo to 20250529124718-87c2dc380cec
Signed-off-by: Maicon Costa <maiconscosta@gmail.com>
* Add owners to pkg/build
Signed-off-by: Maicon Costa <maiconscosta@gmail.com>
* update workspace
Signed-off-by: Maicon Costa <maiconscosta@gmail.com>
* update pkg/build/go.sum
Signed-off-by: Maicon Costa <maiconscosta@gmail.com>
* update pkg/build/go.sum
Signed-off-by: Maicon Costa <maiconscosta@gmail.com>
---------
Signed-off-by: Maicon Costa <maiconscosta@gmail.com>
* Pyroscope: Add annotations frame to series response
* Adapt to API change, add tests
* Run make lint-go
* Fix conflicts after rebase
* Add annotation via a separate data frame
* Process annotations fully at the datasource
* Add mod owner for go-humanize
* Pyroscope: Annotations in Query Response can be optional
---------
Co-authored-by: Piotr Jamróz <pm.jamroz@gmail.com>
* Create libpqToPGX feature toggle
* Refactor PostgreSQL datasource to support PGX with feature toggle
- Updated `ProvideService` to accept feature toggles for enabling PGX.
- Modified integration tests to use the new PGX connection method.
- Introduced new functions for handling PGX connections and queries.
- Enhanced TLS configuration handling for PostgreSQL connections.
- Updated existing tests to ensure compatibility with PGX and new connection methods.
* Update PostgreSQL datasource to enhance connection pooling and error handling
- Increased `MaxOpenConns` to 10 in integration tests for improved connection management.
- Refactored connection handling in `newPostgresPGX` to return a connection pool instead of a single connection.
- Updated health check error handling to utilize context and feature toggles for better error reporting.
- Adjusted `DisposePGX` method to close the connection pool properly.
- Enhanced query execution to acquire connections from the pool, ensuring efficient resource usage.
* Cleanup
* Revert postgres_test unnecessary changes
* Rename feature toggle from `libpqToPGX` to `postgresDSUsePGX`
* Add null check to dispose method
* Fix lint issues
* Refactor connection string generation
* Address comment in health check file
* Rename p to pool
* Refactor executeQueryPGX and split into multiple functions
* Fix lint issues
* The returning error message from PGX is enough no need to separate the error code.
* Move TLS handling to newPostgresPGX function
* Disable ssl for integration tests
* Use MaxIdleConns option
* Remove old feature toggle
* Rename`generateConnectionConfigPGX` to `generateConnectionStringPGX`
* Add back part of the error messages
* Don't show max idle connections option when PGX enabled
* Address comments from Sriram
* Add back Sriram's changes
* PostgreSQL: Rework tls manager to use temporary files instead (#105330)
* Rework tls manager to use temporary files instead
* Lint and test fixes
* Update pkg/tsdb/grafana-postgresql-datasource/postgres.go
Co-authored-by: Ivana Huckova <30407135+ivanahuckova@users.noreply.github.com>
* Update betterer
---------
Co-authored-by: Ivana Huckova <30407135+ivanahuckova@users.noreply.github.com>
---------
Co-authored-by: Ivana Huckova <30407135+ivanahuckova@users.noreply.github.com>
* jaeger backend migration
* move processing to JaegerClient.Search
* fix upload query error source type
* suggestions
* lint
* fix link to traceid query
* fix tests
* fix tests
* use consistent types
* add tests for TransformSearchResponse fn
* test search function
* fix filtering using tags
* suggestion
* remove unnecessary arguments
* use logfmt parser for tags
* test
* test
* use logfmt for query tags
* update
* go fmt
* run backend for all queryTypes
* run make update-workspace
* assign owner to logfmt pkg in go.mod
* apply suggestions
* update tests
* trigger workflows?
* DataData DS: Add Frame type to Random Walk queries
* Time column should not be nullable
Conversion for the SQL (server-side) expression was failing with:
`failed to convert data frames to long format for sql: missing time field`
Because we weren't seeing a match at:
`if f.Type() == data.FieldTypeTime {`]
here:
acb0e6b609/pkg/expr/convert_to_full_long.go (L155)
The DataPlane docs (https://grafana.com/developers/dataplane/timeseries)
state that:
> - The Time field(s):
> - Should have no null values
So making the fix here
This adds more checks on the type of errors returned from InfluxDB
fsql datasource queries. The Grpc error code `Unavailable` is mapped
to a downstream error source.
* Jaeger: Run traceID query through backend with feature toggle enabled
* Reorganize
* Update comments and test
* Fix snapshot test
* Only pass time params if enabled in cofig
* Rename applyVariables to applyTemplateVariables
* Rename transformResponse to transformTraceResponse
* Update comment
* Revert "CloudWatch: Import new grafana-aws-sdk with PDC fix (#103249)"
This reverts commit f2b5b4e0c0230b9cc8553b0c768582698610da00.
* Revert "CloudWatch: Migrate to aws-sdk-go-v2 (#103106)"
This reverts commit a65cc0df93c5568d0ba32f0171a3c2cd04997cf7.
* make update-workspace
* Support PDC for Influx SQL
* Correctly set the gRPC transport creds dial option
* Logging and refactor
* Allow URLs with no scheme
* Use passthrough resolver for socks proxy
* Update comment
* Extract url parsing and add test
* Check err
This adds more checks on the type of errors returned from InfluxDB fsql
datasource queries. Grpc errors with code `InvalidArgument`, `PermissionDenied`
and `NotFound` are mapped to downstream errors.
- Add frame type So SQL expressions knows how to convert it
- No FrameType version so how other expressions do not change how they handle it, and continue to catch the special graphite handling in SSE convert
Fixes#100545
Streaming queries/metrics do not work if TLS is enabled and basic auth is not.
"Save & test" while adding/editing a tempo datasource throw `e.data is undefined` in the ui.
Gafana server logs report:
> logger=grafana-apiserver t=2025-02-12T17:55:29.131036665Z level=info msg="[core] [Channel #42 SubChannel #43]grpc:
> addrConn.createTransport failed to connect to {Addr: \"tempo:3200\", ServerName: \"tempo:3200\", }. Err: connection
> error: desc = \"error reading server preface: read tcp 127.0.0.1:55432->127.0.0.1:3200: read: connection reset by
> peer\""
> logger=grafana-apiserver t=2025-02-12T17:55:36.835523455Z level=info msg="[core] [Channel #31 SubChannel #32]grpc:
> addrConn.createTransport failed to connect to {Addr: \"tempo:3200\", ServerName: > \"tempo:3200\", }. Err: connection
> error: desc = \"error reading server preface: EOF\""
Fix by using TLS when enabled regardless of basic auth settings.
Co-authored-by: André Pereira <adrapereira@gmail.com>
* CI: Allow Bench conversion to fail
We shouldn't mark PRs and commits as X if they fail to convert logs with Bench.
* Fix: Prints should always include new lines
* fix: remove unused import
InfluxDB Flux queries can return a http.Error type that we can use to
assign errorsource to.
This PR checks for that type of error, and uses our errorsource
machinery to assign an errorsource.
This is part of https://github.com/grafana/data-sources/issues/377
* Update dashboard app to use app SDK v0.35.0
What
This change updates dashboard app to use app SDK v0.35.0 and adds new
Makefile target for running codegen for all apps, in opt-in manner.
Currently only dashboards app is opted in.
Additionally, this changes dashboard app Makefile to properly install
and update app SDK versions when generating code, with app SDK version
pinned in the Makefile itself.
Why
The upgrade addresses issues with `DeepCopy` methods, while the Makefile
targets ensure that codegen is easy to run and uses reproducible
environments.
Signed-off-by: Igor Suleymanov <igor.suleymanov@grafana.com>
* Run make update-workspace
Signed-off-by: Igor Suleymanov <igor.suleymanov@grafana.com>
* Fix deepcopy methods
Signed-off-by: Igor Suleymanov <igor.suleymanov@grafana.com>
* Re-run CUE codegen to satisfy the CI
Signed-off-by: Igor Suleymanov <igor.suleymanov@grafana.com>
* Run make update-workspace
Signed-off-by: Igor Suleymanov <igor.suleymanov@grafana.com>
* Update to v0.35.1
Signed-off-by: Igor Suleymanov <igor.suleymanov@grafana.com>
---------
Signed-off-by: Igor Suleymanov <igor.suleymanov@grafana.com>
* Remove aggregate by
* Update betterer
* Run toggles test
* Add back group by in schema so it can be checked for
* Show error if group by
* Update error message
* Allow user to remove the group by from the query
* Fix assertion
---------
Co-authored-by: Piotr Jamróz <pm.jamroz@gmail.com>
