Release notes for September 2026

Grepr released these feature updates, improvements, and bug fixes in September 2026.

September 26, 2026

Other improvements and fixes

Grepr now imports the queries used by your Datadog monitors and dashboards as soon as you save a valid application key, and the Parsed alert queries dialog has an import button to pick up changes without waiting for the next hourly import. See Identify log events that should bypass aggregation.
The field menu in the log details panel now has an Analyze section that fills in the Analyze form to group by a field, count its unique values, count it over time, or add the common numeric aggregations, without running the analysis until you choose to. See Aggregate logs with Analyze mode.
The query search in the Parsed alert queries table now matches queries that contain all of the words you enter, in any order, and highlights the matches; the previous exact-substring search is still available in the search mode menu.
The Parsed alert queries dialog now shows the measured passthrough percentage and its measurement time beside each query, and you can filter the table by passthrough range.
The Datadog integration form now checks your application key as you enter it and after you save it, and lists the permissions it grants and any it is missing.
New Datadog MCP integrations default to Datadog’s supported /v1/mcp endpoint.
The on-premises Helm charts now accept tolerations for the pipeline operator and the Kafka health check, and can read the Kafka health check password from an existing Kubernetes Secret.
In Log Explorer and Trace Explorer, the result count now sits with the results table instead of next to the Search button, so the search controls stay on one row.
The pipeline illustration now has a dedicated dark mode version instead of an inverted copy of the light one.
Fixed an issue where queries in the New Relic log query-like syntax that used lowercase and, or, or not treated them as search terms instead of operators, so the query matched no events.
Fixed an issue where some trace pipelines showed 0 spans out on the pipeline overview even while spans reached their sinks.
Fixed an issue where, when one log event fed more than one reducer in a pipeline, a summary from one reducer could pick up tags or attributes from another.

September 21, 2026

Other improvements and fixes

You can now set the field a Datadog log sink writes the host to with the Host field override setting, for Datadog organizations that removed hostname from their host attributes for JSON log preprocessing.
In Grepr CLI version 1.11.5 and later, job:list omits each pipeline’s job graph unless you pass --include-graph, the list and get commands accept --fields to return only the fields you name, and the GREPR_OUTPUT_FORMAT environment variable sets the default output format. See CLI command reference.
Fixed an issue where the jobs list did not show the link to view a backfill job’s logs in your vendor.
Fixed an issue where the log search tool of the Grepr MCP server failed on some requests.

September 19, 2026

Connect an AI assistant to Grepr with the MCP server

Grepr hosts a remote Model Context Protocol server, so an AI assistant can work with your organization directly: list pipelines, browse datasets and templates, search your logs, validate Flink SQL and Grok rules, and propose pipeline changes. Claude and Claude Code sign in with your own Grepr account over OAuth, Claude in Slack connects through a downloadable plugin, and any other MCP client can authenticate with a service account access token. Every tool call is authorized with the same role-based permissions as the Grepr UI and REST APIs. See Connect to the Grepr MCP server.

Other improvements and fixes

The results grid and the Analyze results table now keep your column order, widths, and sort in the page URL, so a reload or a shared link restores the table as you left it; when an analysis comes back at its row limit, Sort on the server pushes the sort into the query and runs it again. See Aggregate logs with Analyze mode.
An Analyze SELECT column that is neither an aggregate nor one of the grouping expressions now offers a Group by button beside its message, which adds the grouping for you once the server confirms that the amended analysis translates.
Fixed an issue where the billing page failed to load for organizations whose contract does not start on the first of a month.
Fixed an issue where a log send that timed out and returned an error could still be ingested, so the retry from your collector was stored as a duplicate.
Fixed an issue where grouping by a tag in Analyze used only the first of its values, so events carrying the same set of values in a different order landed in different groups; the suggested accessor now reads the whole set as ARRAY_TO_STRING(SORT_ARRAY(tags['service']), ',').
Fixed an issue where a New Relic integration that failed part way through setup was left behind unusable and blocked creating it again under the same name.
Fixed an issue where a pipeline could restart repeatedly instead of recovering after an upgrade timed out.

September 16, 2026

Authenticate OTLP ingestion with a header

You can now send OTLP logs and traces to endpoints that read your Grepr token from an x-grepr-token header instead of carrying it as a segment of the URL path.

Other improvements and fixes

Several improvements were made to Analyze mode in Log Explorer, including checking your analysis with the server as you write it and completion suggestions drawn from the vocabulary the server accepts.
You can now set the records path and the file reading strategy when configuring an S3 file source, so the source can read files that don’t use the default layout.
Log events that carry a capitalized Message attribute now have it remapped to the message field by default.
Fixed an issue where logs in some circumstances were not correctly reporting their severity in Datadog.

September 15, 2026

Improvements to Analyze mode in Log Explorer

Analyze mode now opens from the Analyze… drawer under the search controls rather than from the menu on the Search button, and the drawer spans the width of the page so a long analysis stays readable. The analysis you build is held in the page URL, so you can copy a link to share it and use browser Back and Forward to move between the analyses you built. You can give a grouping expression an alias and reuse that name in conditions and sort order, sort a column by clicking its header, and drag result columns into the order you want, which the CSV export follows. When an expression cannot be computed, the message names your clause and the text you wrote instead of internal column names. See Aggregate logs with Analyze mode and Analyze function reference.

Other improvements and fixes

Log Explorer now shows an advisory under the search box when a query scans more data than it needs to, such as a search across every log field or a search with no service or host filter.
Exceptions suggested from your Datadog dashboards now include the log queries in issue stream, log stream, split and grouped, and legacy top list widgets.
You can now copy a pipeline’s JSON configuration to your clipboard or download it as a file from the Pipeline JSON menu.
Fixed an issue where moving a pipeline between Quickstart and Advanced Configuration lost the data lake dataset, leaving the pipeline unsaveable and reporting that the data warehouse was not found.
Fixed an issue where the SQL draft preview showed none of the filters for a pipeline built from a raw job graph.
Fixed an issue where a Datadog-like or New Relic log query-like query that used a quoted field name, such as "AcmeStoreId":87971, failed to parse in Log Explorer, in pipeline filters, and when importing exceptions from a Datadog dashboard.
Fixed an issue where importing exceptions from New Relic failed for a monitor or dashboard whose query combined FACET, SINCE, and a trailing LIMIT.
Fixed an issue where reopening the settings for a pipeline step showed blank or stale values instead of your saved configuration.

September 6, 2026

Aggregate your logs with Analyze mode in Log Explorer

You can now run aggregation queries over your logs directly in Log Explorer. Open the menu on the Search button and select Analyze, then describe what to compute as Grepr SQL expressions, such as counts, sums, and percentiles, and group the results by any field. You can bucket the results into time windows to see how the aggregates change over the time range, and the builder autocompletes function names and shows their signatures and examples as you type. See Aggregate logs with Analyze mode.

Reach Grepr support from inside the app

You can now start a conversation with Grepr support from any page in the app using the in-app messenger. Your messages are tied to your organization account, so support has the context to help you.

Other improvements and fixes

You can now preview how a masking step transforms sample log lines before you apply it to a pipeline.
Managing exceptions in a logs pipeline is easier, with search, ID-column filtering, and bulk selection that applies only to the rows your search returned.
You can now add a field shown in the Log Explorer details panel as a column in the results table using the Add as column option.
You can now download a pipeline’s draft job graph as a JSON file from the pipeline editor without applying it.
The Grepr plugin now includes a sql-analytics skill for running read-only analytical SQL queries over a dataset, available after you update the plugin.
Improved the performance and memory usage of trace pipelines that forward to Datadog.
The data lake step for agent session pipelines now shows the dataset name with quick links to open it in the explorer or edit its settings, matching the log and trace steps.
The data explorer button on an agent session pipeline now opens the agent sessions page with the matching dataset and time range.
Fixed an issue where the pipeline details page showed default reducer settings instead of your saved reduction window, new-pattern signal, and nested group-by settings.
Fixed an issue where setting up a pipeline could hang instead of completing when the underlying cluster was slow to schedule its health checks.
Fixed an issue where the CLI job:draft command failed on pipelines whose vendor sink feeds another sink.
Fixed an issue where importing a Datadog dashboard failed to parse when a template variable defined an out-of-range bound.