This reverts commit 046791e2be.
Is causing malformed frames, and resulting in errors about mismatched field lengths in the logs, No data in the UI, and will cause SSE to panic.
I think this is because of the global slice vars since it seems to take concurrency to replicate it.
When running in dev mode, error messages would contain an additional "error" property alongside "message". Since this causes confusion, that has been removed and now error messages are the same both modes (using "message").
Changes SSE to not always fail all queries when one fails. Now only the query itself, and nodes that depend on it will error.
---------
Co-authored-by: Gilles De Mey <gilles.de.mey@gmail.com>
- The util/converter Prometheus response json parse was not checking for errors while parsing. It now does. In particular, if `[dataproxy]/response_limit` is set in Grafana's config, it will now recognize the limit error.
- Fixes#73747
- Adds `jsonitere` package, which wraps json-iterator/go's Iterator's Methods with methods that return errors, so errcheck linting can be relied upon
- Impact:
- If something was sending malformed JSON to the prometheus or loki datasources, the previous code might have accepted that and partially processed the data
- Before there may have been partial data with no error, where as no there may be errors but they will have no partial results, just the error.
Execute all queries to the same datasource in a single request.
Uses the query index and the graph node ID index, and then a stable dependency graph sort based on node input index number in attempt to keep the original query order intact.
* Azure: AzureMonitorMetrics - change response to be dataplane compliant
can be disabled via toggle azureMonitorDataplane
Co-authored-by: Andreas Christou <andreas.christou@grafana.com>
Currently does not do anything as it is for use with a future enterprise PR
Eventually: "Enables writing multiple items from a single query within Recorded Queries"
Takes a specific code path for data that identifies itself as dataplane instead of "guessing" what the data is.
The data must identify itself by being in the dataplane by having both the following frame metadata properties:
- TypeVersion property that is greater than 0.0
- 'Type' property
The flag is disableSSEDataplane and disables this functionality and uses the old code for all queries regardless.
See https://github.com/grafana/grafana-plugin-sdk-go/blob/main/data/contract_docs/contract.md for dataplane details.
SDK release notes:
* Logs contract: ignore remaining fields by @gabor in #659
* Logs contract: more robust field finding, explicit approach by @gabor in #660
* Tracing: Support multiple OTel propagators by @xnyo in #663
* Tracing: Add more details to HTTP Outgoing Request by @xnyo in #664
* Data: Encode Nanosecond into JSON by @kylebrandt in #647
* Data: cmp tests using FrameTestCompareOptions() will no longer ignore time differences beyond millisecond resolution
Use latest github.com/grafana/grafana-plugin-sdk-go which includes changes to the TypeVersion property (always present in JSON).
Also included is squtil changes: SQL util - allow using the database scan type for converters
* Start work on allowing certain resources to pass through Cache-Control headers.
---------
Co-authored-by: Marcus Efraimsson <marcus.efraimsson@gmail.com>
This restores the FromAlert header to prometheus for Grafana managed alert Queries.
It does this by reverting "Prometheus: Remove middleware for custom headers (#51518)" , but also changing it so it is only the FromAlert header.
This reverts commit 2372501368.
When there is a single frame with no fields (e.g. splunk datasource) SSE errors when trying to figure out the data type. This frame needs to exist since this is where the executedQueryString metadata exists.
This adds a new return type to SSE to represent no data, so the original frame with its metadata can still be maintained.
* Chore: Remove deprecated no-op option PreferServerCipherSuites
from the tls configs.
Per golang docs: "PreferServerCipherSuites is a legacy field and has no effect."
* Alerting: Allow __value__ label in notifications
was being removed by removePrivateItems
discoverd in #36020, but issue is not about that specifically
* __value__ label to __value_string__ annotation
and .ValueString extended property for notifications
When using mulit-dimensional Grafana managed alerts (e.g. SSE math) extract refIds values and labels so they can be shown in the notification and dashboards.
removes "rpc error: code = InvalidArgument desc =" like strings from error messages.
This came from old grpc stuff left from when SSE was a plugin (that probably should
not have been used even when it was a plugin)
Rules/notifications/etc migration will now be activated with feature flag alone.
When the feature flag is enabled dashboard alerts are migrated into the system.
When the feature flag is removed, all migrated and newly created alerts in the new system are deleted.
When, and currently only when using a classic condition, evaluation information is added (which is like the EvalMatches from dashboard alerting).
This is returned via the API and can be included in notifications by reading the `__value__` label attached `.Alerts` in the template. It is a string.
Instead put in package folder but with package name suffixed with _test
This enables code coverage within the pkg while still allow the tests to operate from external to package perspective (only exported things).
makes it so the feature flag can be turned on off, and the migration will be cleared and rerun. All existing NG alert rules, configuration settings, etc are removed when disabling the feature flag.
for https://github.com/grafana/alerting-squad/issues/142
Co-authored-by: Sofia Papagiannaki <sofia@grafana.com>
Kyle Brandt
