What parseablehq/parseable shipped
Written by FoxPlug from the repository's own releases; not affiliated with Parseable. Paste your own repository in the box below and FoxPlug writes the same weekly update for your product. This page is an automatic summary of the releases, pull requests and commits in github.com/parseablehq/parseable; Parseable did not write it and does not use or endorse FoxPlug. Every line links to the change it describes.
Get a weekly update like this for your product, free
Week of October 5, 2026
What shipped
- Added a REST API for an external server that discovers available tools and calls supported data, query, alert, trace, identity and saved-object operations, with argument validation and access controls. Pull request #1816
- External tool requests now go through query authorization before running, and listings and calls are limited to read-only and expensive-read tools. Pull request #1817
- Fixed failed alert evaluation so a successful retry restores the configured evaluation frequency and the task is no longer stopped entirely by consecutive failures. Pull request #1814
- Deleting a tenant now stops its active processing and clears its cached runtime state. Pull request #1806
- Added an endpoint for listing available tools and their input schemas, requiring query permission, and abstracted internal processing out of handler functions across alert, cluster, health, stream, query, role, target, trace, dashboard, user and filter operations. Pull request #1804
- Released version 3.2.5. Release
- Process metrics now report a CPU limit alongside the existing process CPU usage metric, read from cgroup v2 with cgroup v1 and logical CPU fallbacks. Pull request #1796
- The home view now includes a setup checklist showing whether data has been ingested, a Keystone or additional user has been created, and alerts are configured for the current user. Pull request #1811
- Trace searches and individual trace lookups now require one dataset name per request instead of allowing multiple dataset names. Pull request #1819
- Updated the bundled UI asset for release 3.2.5. Pull request #1818
Why it matters
This week adds read-only external tool access with query authorization and argument validation, plus an endpoint listing available tools and their input schemas, so external integrations can be scoped to safe operations. Alert evaluation failures no longer silently stop a task, and tenant deletion now stops that tenant's processing instead of leaving tasks running. Process CPU limit reporting and a home setup checklist also land, and version 3.2.5 ships these changes.
Changelog entry
- Release v3.2.5. Release
- Added a REST API for discovering available tools and calling supported data, query, alert, trace, identity and saved-object operations, with argument validation and access controls. Pull request #1816
- External tool requests now undergo query authorization, and listings and calls are limited to read-only and expensive-read tools. Pull request #1817
- Added an endpoint for listing available tools and their input schemas, and abstracted internal processing from handler functions across alert, cluster, health, stream, query, role, target, trace, dashboard, user and filter operations. Pull request #1804
- Fixed failed alert evaluation so successful retry restores the configured evaluation frequency and consecutive failures no longer stop the task entirely. Pull request #1814
- Deleting a tenant now stops its active processing and clears its cached runtime state. Pull request #1806
- Process metrics now report
parseable_process_cpu_limit_coresalongside the existing CPU usage metric, read from cgroup v2 with cgroup v1 and logical CPU fallbacks. Pull request #1796 - The home view now includes a setup checklist for ingested data, a created Keystone or additional user, and configured alerts. Pull request #1811
- Removed unused code, including overview API handlers and metastore overview methods; the home screen now shows datasets only. Pull request #1810
- Trace searches and individual trace lookups now require a single dataset name per request. Pull request #1819
- Updated the bundled UI asset for release v3.2.5. Pull request #1818
This week: read-only external tool access with query authorization, an endpoint listing tools and their input schemas, a fix so failed alert evaluation no longer stops the task, tenant deletion now stops its processing, a process CPU limit metric, and version 3.2.5.
This week in Parseable: external tool access is now limited to read-only and expensive-read tools, and every external tool request goes through query authorization before it runs, with argument validation on calls. An endpoint now lists available tools and their input schemas. Failed alert evaluation is fixed so a successful retry restores the configured evaluation frequency instead of stopping the task. Deleting a tenant now stops its active processing and clears its cached state. Process metrics report a CPU limit alongside CPU usage. The home view adds a setup checklist, trace lookups require a single dataset name, and version 3.2.5 is released.
Week of September 28, 2026
What shipped
- Release v3.2.4 makes the series-hashing function public for read-write use and fixes query execution when the receiver drops. Release
- Release v3.2.3 fixes query execution during hot-tier eviction, exposes node counts for analytics, and adds slowest query interval tracking. Release
- The
compute_series_hashfunction is now publicly accessible for integrations working with flattened metric series. Pull request #1801 - Query execution now aborts properly when the receiver drops, preventing resource waste. Pull request #1802
- Slowest query duration is now tracked and exported as
parseable_slowest_query_duration_secondswithtime_range_secondsandquery_typelabels. Pull request #1795 - Queries now take a guard during hot-tier eviction to prevent data loss and skip unnecessary sync cycles. Pull request #1797
- Active and inactive ingestor and querier counts are now exposed for enterprise analytics to reuse existing cluster liveness logic. Pull request #1789
- Query execution-time bucket boundaries are now reusable across SQL and Enterprise PromQL metrics. Pull request #1799
Why it matters
Two patch releases this week address critical query execution issues and improve observability. v3.2.4 and v3.2.3 fix hot-tier data safety and receiver handling, while adding metrics that help operators understand query performance patterns and cluster health.
Changelog entry
- Make fn
compute_series_hashpublic for rw use Release - Fix: abort query exec when receiver drops Pull request #1802
- Fix: query execution at the time of hottier eviction Release
- Expose node counts for enterprise analytics Pull request #1789
- Feat: add slowest query interval metric Pull request #1795
- Refactor: remove slow query metric and expose query buckets Pull request #1799
V3.2.4 and v3.2.3 shipped. Query execution is more reliable during hot-tier eviction, receiver drops are handled properly, and new metrics expose slowest query duration and node counts.
Two releases this week: v3.2.4 and v3.2.3. Improvements include proper query execution during hot-tier eviction to prevent data loss, correct handling of receiver drops, and new observability metrics for slowest query duration and cluster node counts. The series-hash function is now public for integrations.
Week of September 21, 2026
What shipped
- Parseable v3.2.2 released with bug fixes for ingestion and OpenTelemetry JSON handling, plus new disk usage and process memory limit metrics. Release
- Disk usage metrics added to track staging and Hot Tier storage consumption with
parseable_disk_used_bytesandparseable_disk_total_bytesmetrics. Pull request #1791 - Process memory limit metric exposed as a Prometheus metric to enable memory utilization calculations. Pull request #1790
- Conversion and object store sync operations now tracked with histogram metrics including phase information. Pull request #1793
- Retention cleanup tasks now only target live ingestor nodes, removing unnecessary requests to other server modes. Pull request #1792
- Dataset field count limit restriction removed while maintaining 1000 as default, and RFC 2822 timestamps handled as UTF8 to prevent JSON decoding failures. Pull request #1788
- Docker images now published with both version-specific tags and latest tag to Quay.io at each release. Pull request #1787
Why it matters
This release improves observability with new metrics for disk usage and memory limits, making it easier to monitor resource consumption. Critical bugs around data ingestion and retention task routing have been fixed, reducing errors and unnecessary operations.
Changelog entry
- Added disk usage metrics for staging and Hot Tier storage Release
- Added process memory limit metric to Prometheus registry Pull request #1790
- Added
conversion_task_durationandobject_store_sync_durationhistogram metrics Pull request #1793 - Removed dataset field count limit restriction, improved JSON timestamp handling Pull request #1788
- Fixed retention cleanup to only target ingestor nodes Pull request #1792
- Fixed RFC 2822 timestamp decoding by keeping as UTF8 Pull request #1788
- Fixed shutdown sync operations to skip Query Server mode Pull request #1793
- Docker images now pushed with latest tag at every release Pull request #1787
Parseable v3.2.2 is out. Better observability with disk and memory metrics, fixes for ingestion and retention tasks, plus improved JSON handling. Get it now.
We've released Parseable v3.2.2 with significant improvements to observability and reliability. New disk usage and process memory metrics give you better visibility into resource consumption. We've also fixed critical bugs in data ingestion and retention task routing, and improved handling of JSON and RFC 2822 timestamps. Docker images now include both version-specific and latest tags for easier updates.