What Security-Onion-Solutions/securityonion shipped
Written by FoxPlug from the repository's own releases; not affiliated with Securityonion. 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/Security-Onion-Solutions/securityonion; Securityonion 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 November 2, 2026
What shipped
- Released version 3.4.0-20261102 with the download and verification instructions for the ISO. Release
- Changed so-elasticsearch-ilm-policy-load output so it now surfaces ILM policy errors. Pull request #16378
- Added installation of securityonion-so-crates on Security Onion Desktop. Pull request #16377
- Removed the x86-64-v3 CPU pre-flight check from soup and the --skip-cpu-check flag; soup --skip-cpu-check now prints usage and exits 1. Pull request #16373
- Changed the ISO build to skip a leftover registry.tar when tags do not match SOVERSION. Pull request #16374
Why it matters
This week brings the 3.4.0 release to users, along with fixes that make failures easier to see: ILM policy errors now appear in the policy load output, and stale ISO registry artifacts are skipped when tags do not match. The soup CPU pre-flight check was removed because the underlying image issue is fixed, so one flag and its related check are gone.
Changelog entry
- Release 3.4.0-20261102 Release
- Surface ILM policy errors in so-elasticsearch-ilm-policy-load output Pull request #16378
- Install securityonion-so-crates on Security Onion Desktop Pull request #16377
- Remove x86-64-v3 CPU pre-flight check and --skip-cpu-check flag from soup Pull request #16373
- Skip leftover ISO registry.tar when tags do not match SOVERSION Pull request #16374
Security Onion 3.4.0 released. ILM policy errors now show in so-elasticsearch-ilm-policy-load output, securityonion-so-crates installs on Security Onion Desktop, stale ISO registry.tar is skipped on tag mismatch, and soup no longer has the CPU pre-flight check or --skip-cpu-check
Security Onion 3.4.0 is out. In this week's work, so-elasticsearch-ilm-policy-load now surfaces ILM policy errors, and securityonion-so-crates is installed on Security Onion Desktop. ISO builds skip a leftover registry.tar when tags do not match SOVERSION. The x86-64-v3 CPU pre-flight check and the --skip-cpu-check flag were removed from soup since the image issue is fixed, so soup --skip-cpu-check now prints usage and exits 1.
Week of October 26, 2026
What shipped
- All containers now use the host CA trust store together with the grid CA instead of the roots their images shipped with and the stale so-catrust copies. Pull request #16357
- Removed the unused elasticsearch.p12 and sokeys keystore, which were left over from a removed transport-TLS plugin. Pull request #16354
- Backported the salt 3006.29 eMudhra and emSign CA fix into the 3006.19 onedir build. Pull request #16350
- Added configurable timeouts that limit how long each detection engine sync can run after detections are modified. Pull request #16358
- Added
custom_bind_mountssupport to so-kafka and so-soc so the setting is no longer silently ignored for those containers. Pull request #16356 - Fixed false positive scan results. Pull request #16351
- Updated analyzer dependencies. Pull request #16341
- Switched analyzer dependencies to the Python 3.14 build. Pull request #16361
Why it matters
This week changes how containers trust certificates, moving them to a shared host trust store and grid CA, and removes stale keystore material. Detection engine syncs can now be bounded by configurable timeouts, and the certificate authority fix for eMudhra and emSign is carried into the salt build in use.
Changelog entry
- Use the host CA trust store, with the grid CA, for all containers. Pull request #16357
- Remove unused elasticsearch.p12 and sokeys keystore. Pull request #16354
- Backport salt 3006.29 eMudhra/emSign CA fix into the 3006.19 onedir. Pull request #16350
- Add configurable timeouts for detection engine syncs. Pull request #16358
- Add
custom_bind_mountsto so-kafka and so-soc. Pull request #16356 - Fix false positive scan results. Pull request #16351
- Update analyzer dependencies. Pull request #16341
- Switch analyzer dependencies to Python 3.14. Pull request #16361
This week: all containers now use the host CA trust store with the grid CA, detection engine syncs get configurable timeouts, and the eMudhra/emSign CA fix is backported to the salt 3006.19 onedir.
This week's changes to the project: containers now use the host CA trust store together with the grid CA, replacing the roots images shipped with and the stale so-catrust copies. The unused elasticsearch.p12 and sokeys keystore were removed. Each detection engine now has configurable timeouts for how long a sync may run after detections change. The salt 3006.29 eMudhra and emSign CA fix was backported into the 3006.19 onedir build. custom_bind_mounts now works for so-kafka and so-soc, false positive scan results were fixed, and analyzer dependencies were updated, including a switch to Python 3.14.
Week of October 12, 2026
What shipped
- Alerts queries added to pillar from the SOC config UI now merge with default queries by name instead of replacing them, so defaults added later, such as the licensed Group By Assessment, Name query, still reach soc.json. Pull request #16328
- Added a licensed Group By Assessment, Name alerts query that groups by the alert triage assessment, rule name, severity label and rule uuid. Pull request #16302
- Removed the Notifier agent mapping because AlertTriage now sends its own notification through the Notify skill, and dropped the Notifier annotation while rewording AlertTriage's description. Pull request #16323
- Renamed the readonly setting to readonlyUi in the YAML configuration. Pull request #16332
- Added a readonly attribute to the earliest alert time field. Pull request #16331
- Combined default and availableModels entries added via pillar. Pull request #16322
- Simplified the automation turn timeout setting to turnTimeoutSeconds on automationSettings. Pull request #16333
- Remapped the enterprisesearch integration components. Pull request #16319
Why it matters
Several changes this week affect how SOC configuration written through the interface interacts with shipped defaults, particularly for alerts queries and model listings. The new licensed Group By Assessment, Name query gives a view grouped by triage assessment for alerts that have been assessed. Notification handling for alert triage also moved into AlertTriage itself with the Notifier agent mapping removed.
Changelog entry
- Combine default and alerts queries added via pillar by name, keeping pillar order and edits so later defaults reach soc.json. Pull request #16328
- Add a licensed Group By Assessment, Name alerts query grouping by the alert triage assessment, rule name, severity label and rule uuid. Pull request #16302
- Remove the Notifier agent mapping and annotation since AlertTriage sends its own notification through the Notify skill, and reword AlertTriage's description. Pull request #16323
- Change readonly to readonlyUi in the YAML configuration. Pull request #16332
- Add a readonly attribute to the earliest alert time field. Pull request #16331
- Combine default and availableModels entries added via pillar. Pull request #16322
- Simplify the automation turn timeout setting to turnTimeoutSeconds on automationSettings. Pull request #16333
- Remap enterprisesearch integration components. Pull request #16319
This week: alerts queries from the SOC config UI now merge with defaults by name, so defaults added later still reach soc.json. Added a licensed Group By Assessment, Name query. AlertTriage sends its own notifications, so the Notifier agent mapping is gone.
This week's changes for SOC configuration and alert triage. Queries added to defaults via pillar are now merged with alerts queries saved from the SOC config UI by name, keeping pillar order and edits, so defaults added later such as the licensed Group By Assessment, Name query reach soc.json. A licensed Group By Assessment, Name alerts query was added, grouping by the alert triage assessment, rule name, severity label and rule uuid; untriaged alerts are not in this view. AlertTriage now sends its own notification through the Notify skill, so the Notifier agent mapping and annotation were removed and AlertTriage's description reworded. Also, readonly became readonlyUi in the YAML configuration, the earliest alert time field gained a readonly attribute, default and availableModels entries a
Week of October 5, 2026
What shipped
- Initial Correlations support added with Sigma Correlation rules, a SOC migration that fills rule type on existing detections, and ElastAlert defaults that raise
old_query_limitfrom 5 minutes to 24 hours. Pull request #16269 - Sigma pipelines moved into a managed directory at /opt/so/conf/soc/sigma_pipelines/, mounted read-only into SOC, replacing the two single-file mounts and writing the final pipeline as
sigma_final_pipeline.yml. Pull request #16287 - New allowExternalMarkdownImages SOC setting added, default off, to let user-written Markdown such as case comments load external images. Pull request #16312
- Evaluation changed to use Suricata for metadata. Pull request #16317
- Elasticsearch updated to 9.4.8. Pull request #16306
- New destination timeout annotation added and a false positive fixed. Pull request #16310
- Elastic Agent installer regenerated. Pull request #16311
- Reverted the auto apply state queue fix. Pull request #16297
- Tick interval description updated. Pull request #16300
Why it matters
This week brings initial Correlations support with Sigma Correlation rules and a SOC migration, plus ElastAlert defaults that raise old_query_limit so rules resume after an outage. Sigma pipelines are now managed in a single directory, and a new setting controls whether user-written Markdown can load external images. These changes affect how detections, pipelines, and Markdown content behave for people using the project.
Changelog entry
- Add initial Correlations support with Sigma Correlation rules, a SOC migration for rule type on existing detections, and ElastAlert
old_query_limitraised to 24 hours. Pull request #16269 - Move Sigma pipelines into a managed directory mounted read-only into SOC, replacing two single-file mounts. Pull request #16287
- Add allowExternalMarkdownImages SOC setting, default off, for user-written Markdown external images. Pull request #16312
- Change evaluation to use Suricata for metadata. Pull request #16317
- Update Elasticsearch to 9.4.8. Pull request #16306
- Add new destination timeout annotation and fix a false positive. Pull request #16310
- Regenerate the elastic agent installer. Pull request #16311
- Revert the auto apply state queue fix. Pull request #16297
- Update tick interval description. Pull request #16300
Initial Correlations support with Sigma Correlation rules, Sigma pipelines moved to a managed directory, and a new setting for external Markdown images. Elasticsearch 9.4.8 and an evaluation change to use Suricata for metadata.
This week shipped initial Correlations support with Sigma Correlation rules, including a SOC migration that fills rule type on existing detections and ElastAlert defaults raising old_query_limit from 5 minutes to 24 hours so rules resume where they stopped after an outage. Sigma pipelines are now in a managed directory at /opt/so/conf/soc/sigma_pipelines/, mounted read-only into SOC, replacing two single-file mounts. A new allowExternalMarkdownImages setting, default off, lets user-written Markdown such as case comments load external images. Also included: Elasticsearch 9.4.8, an evaluation change to use Suricata for metadata, a new destination timeout annotation, an elastic agent installer regeneration, and a revert of the auto apply state queue fix.
Week of September 28, 2026
What shipped
- EVTX imports now use the same data stream and parsing as live Windows data instead of a separate import pipeline. Pull request #16278
- Sigma detections on Sysmon and 4688 sources now query plain fields instead of.caseless to fix missing alerts and broken playbooks. Pull request #16279
- Automations are now unified, stored in the database, and marked as advanced and read-only instead of using templates. Pull request #16280
- Removed docker socket access from so-telegraf to prevent container escape risks on all node types. Pull request #16261
- Docker upgraded to 29.8.1 and containerd upgraded to 2.3.6. Pull request #16286
- Service accounts now use /sbin/nologin instead of /bin/false for shell access. Pull request #16263
- New properties added to the Automation configuration system. Pull request #16270
- Empty YAML files are now supported in configuration parsing. Pull request #16277
- Startup errors resolved. Pull request #16267
- Metric alarms and NTF annotations expanded. Pull request #16264
Why it matters
Windows event imports and Sigma detections received fixes that improve data consistency and detection accuracy across different event sources. Security hardening removed container escape vectors from telegraf, and automation handling was refactored for better maintainability.
Changelog entry
- EVTX imports routed to logs- -import data stream with Winlogbeat format conversion and import.evtx pipeline Pull request #16278
- Sigma detections query plain fields instead of.caseless for Defend-independent sources Pull request #16279
- Automations marked as advanced, read-only, and stored in database instead of templates Pull request #16280
- Removed /var/run/docker.sock mounting from so-telegraf container Pull request #16261
- Docker upgraded to 29.8.1 and containerd to 2.3.6 Pull request #16286
- Service accounts configured with /sbin/nologin shell Pull request #16263
- New properties added to Automation configuration Pull request #16270
- Support added for empty YAML file parsing Pull request #16277
- Startup errors resolved Pull request #16267
- Metric alarms and NTF annotations enhanced Pull request #16264
Security-Onion 3 updates: EVTX imports unified with live Windows parsing, Sigma detections fixed for Sysmon/4688, docker socket removed from telegraf, automations refactored to DB storage.
Security-Onion 3 updates improve Windows event handling and detection reliability. EVTX imports now use consistent parsing with live Windows data. Sigma detections fixed for Sysmon and 4688 sources. Security hardening removed unnecessary docker socket access from telegraf. Automations moved to database storage for better management.
Week of September 14, 2026
What shipped
- Isolated the Kratos and Hydra admin API on a dedicated soauth docker network to prevent unauthorized access via host network exposure. Pull request #16221
- Fixed privilege escalation vulnerability where unprivileged service UIDs owned /usr/sbin management scripts executed by root. Pull request #16250
- Fixed code execution vulnerability where so-minion exported unescaped install variables into Salt pillar rendering. Pull request #16246
- Fixed SOC user creation failing when passwords met requirements, with error messages now properly surfaced instead of hidden in logs. Pull request #16255
- Fixed SOC config screen rendering failure caused by docker.networks setting name conflict introduced in the previous release. Pull request #16253
- Fixed so-sensor-clean runaway loop that could fill logs with 1.1 GB in five hours by adding progress checks and pass limits. Pull request #16243
- Corrected Salt-installed management script ownership to prevent privilege escalation from service UIDs to root. Pull request #16250
- Fixed suricata.fileinfo boolean handling to properly represent missing file bytes instead of incorrect long values. Pull request #16249
- Added case-insensitive matching for ESQL and overrode default ESQL runtime parameters. Pull request #16247
- Fixed Zeek shutdown to terminate gracefully and preserve post-termination logs. Pull request #16242
Why it matters
This week addresses several security vulnerabilities including network isolation of authentication APIs, privilege escalation bugs, and code injection risks. A regression from the previous release broke user creation and config rendering, both now fixed. Performance and operational stability improvements prevent cleanup loops and log runaway issues.
Changelog entry
- Isolate Kratos and Hydra admin API on dedicated soauth docker network Pull request #16221
- Fix ownership of /usr/sbin management scripts and Salt default tree to prevent privilege escalation Pull request #16250
- Fix so-minion pillar injection vulnerability and disk.usage handling Pull request #16246
- Fix SOC user creation with valid passwords and improve error messaging Pull request #16255
- Fix config tree error from docker.networks setting conflict Pull request #16253
- Prevent so-sensor-clean runaway loop and concurrent instances Pull request #16243
- Fix suricata.fileinfo boolean gap handling in file.bytes.missing Pull request #16249
- Add case-insensitive ESQL matching and override default runtime parameters Pull request #16247
- Stop Zeek gracefully to preserve post-terminate logs Pull request #16242
This week: fixed three security vulnerabilities in network isolation, privilege escalation, and pillar rendering. Resolved regressions in user creation and config screens. Improved sensor cleanup and Zeek shutdown behavior.
This week's updates address critical security improvements and operational stability. Fixed network isolation for authentication APIs, patched privilege escalation vulnerabilities in management scripts and pillar rendering, and resolved user-facing regressions in account creation and configuration. Additional improvements prevent runaway cleanup processes and ensure Zeek terminates gracefully.