nano SIEM
Releases

nano v2.0.18

Supported in:Self-hosted

nano v2.0.18 verifies every online license check-in, stops parser publish rollbacks on slow Vector reloads and shows OCSF fields in alerts and asset views.

v2.0.18 includes everything in v2.0.17. It is not flagged as a security release.

To upgrade, follow Upgrade a self-hosted deployment. On Docker Compose, scripts/nano upgrade upgrades v2.0.17 to v2.0.18 directly. Earlier v2.x releases also upgrade directly; there are no required stops. nano Self-run installs: read Licensing before you upgrade.

Licensing

nano Self-run. An install now accepts an online license check-in only when the answer is signed by nano. Answers that aren't signed, or that don't match the install, are refused. Signed answers are checked again each time the license is read.

What this means when you upgrade:

  • Licensed features unlock at the first signed check-in. The install checks in at startup, so this is normally a few seconds after it starts. Until then licensed features are locked. If the check-in fails, it retries every 5 minutes.
  • The install must reach the license server. Allow outbound access to api.nano.rs, directly or through your corporate proxy. See Network access.
  • The host clock must be right. A check-in is refused when the host clock is more than 5 minutes from real time. Keep NTP running on every host.
  • Stale licenses still lock after 14 days. If check-ins keep failing, the install keeps its last verified license for 14 days from when nano signed it, then locks.
  • Locking never touches your data. While locked, open-core features and all stored data stay available.
  • Offline and air-gapped licenses are unaffected. They make no online check-in.

Fixes and improvements

Parsers and log sources

  • First OCSF publish no longer fails with 409. Publishing a new OCSF parser for the first time conflicted with its own sample check and returned 409. A real concurrent edit still returns 409.
  • Slow Vector reloads no longer roll back a good config. With many parsers deployed, a Vector reload can take longer than 45 seconds. The deploy used to give up and roll back, and the rollback's own reload then failed the next publish too. The deploy now keeps waiting while Vector is still reloading. It only fails when Vector goes quiet, or when it hits a hard ceiling.
    • VECTOR_RELOAD_ACK_MAX_SECS (default 600) sets that ceiling.
    • VECTOR_RELOAD_ACK_TIMEOUT_SECS keeps its meaning: 45 seconds by default, 0 turns the check off.
  • Parser test lab shows nested OCSF fields. Results list every field as a dotted path, such as http_request.url.hostname and src_endpoint.ip. Long values wrap. unmapped.* fields are grouped under Unmapped Fields.
  • HEC-fed sources show their real state. They no longer read "PULL PAUSED" or "not deployed". The HEC listener is always on. Its row shows receiving or idle from actual arrivals. Other listeners show as push, not pull.
  • Save VRL keeps a live source deployed. Saving a draft no longer flips a published source to "Ready / not deployed". Vector keeps running the published version. The Deploy dialog says it redeploys the published version, and that draft changes go live only with Publish.
  • The VRL editor keeps unsaved edits. It warns before you leave with unsaved changes. After a reload or navigating away, it offers Restore edits or Discard. A draft is never offered over a newer save.

Alerts and detections

  • Alert detail shows OCSF fields. Expanded events and the preview now show OCSF fields such as process.name. They were hidden before.
  • Real-time OCSF alerts carry the full OCSF event. Before, the matched event, finding message and observables used UDM names. They now match what scheduled alerts store. Alerts are larger as a result.
  • Real-time rules record their matches. Live and Alerting real-time rules now record each match, like scheduled rules. "Show me every time this rule fired" now includes real-time rules.
  • Assignment shows names. The Assignee column, the Assign button, the Triage card and the Timeline show the user's name instead of an ID. The Triage card and the Timeline also show who assigned the alert and when.
  • Rule tester counts each event once. Overlapping lookback windows counted the same event several times, so 3 events could show as 9 matches. Each event now counts once, and the sample has no duplicates. Events newer than one schedule tick are now included.
  • New rules get version 1. Every new rule records a v1 on create. The version panel refreshes after save and shows the change reason.
  • Seconds-field schedules read correctly. A 6- or 7-field cron such as */30 * * * * * reads "every 30s", not "every 30m".
  • New rule template uses OCSF fields. It filters on metadata.log_source and scores actor.user.name, and runs on a schedule every 5 minutes. On a new draft the save button reads Create rule.

Asset view

  • Asset dossier works on OCSF. Every OCSF entity's dossier failed with a 400 because of the DNS card. It now loads.
  • Alert pivots find alerts. Open sequence and the alert lane searched a source with no rows. They now find OCSF Findings (category_uid=2), or findings and signal events on UDM.
  • Pivots keep the field you clicked. On OCSF, Open asset view keeps the clicked field (for example device.hostname) and is not offered for fields it can't start from. Investigate searches every OCSF host field. A full hostname is no longer cut to the short name.
  • Empty process trees say why. When events match but none carry the process link ID, the tree says so. For example, "None of the 29 matching events carry process.uid".
  • table keeps the column names you ask for. table time, metadata.log_source dropped both columns, as did UDM names such as src_ip on OCSF. They now come back under the requested names, and later stats, sort, where, timechart and dedup stages read them. See Upgrade notes.
  • Saved searches reopen on their time range. Before, a saved search ran with whatever the picker showed. A search saved on a preset such as "Last 24 hours" now reopens on that preset, relative to now. One saved on a fixed window reopens on that window. The Save dialog shows the range being saved.
  • Multi-year histograms draw correctly. On a window spanning years, the Event Stream histogram piled every year into the first one. Long windows now step by month or year, and labels show the year when the window crosses one.

Access control and audit

  • Audit search covers every event. The audit log search box filtered only the 50 rows on screen. It now searches message, action, user, IP and details for every event in the selected time range. Export all keeps the search, and the export's audit entry records it.
  • More audit detail.
    • API key create records the scopes, expiry and rate limit. Update records each changed field, old and new. Reset and delete record a snapshot of the key, without secrets.
    • Search settings changes record each changed field, old and new.
    • The audit CSV export has a details column.
    • The Action filter lists every action that appears in the log.
  • Fewer permission errors for non-admin roles. The app no longer requests data a role can't read. Home cards, status bar items and Settings controls the role can't read are hidden, instead of showing errors or "No data yet". Server permissions are unchanged.
  • Access control with partial permissions. Roles load without users:view; member counts are hidden. Groups say that viewing members needs users:view, instead of "No members yet".
  • Getting Started is for admins. Users without settings:view are no longer sent into instance setup. A failed status check shows "Status unavailable", not "No sources yet".
  • Clearer license messages. Non-admins see "Ask your administrator" instead of import and billing buttons. The refusal page names the feature, such as "Cases is not licensed". License is in Settings on every Self-run build, and stays reachable while unlicensed. pivt says when it needs a license.

Upgrade notes

  • Licensed installs need a signed check-in. See Licensing. Make sure the install can reach api.nano.rs and the host clock is correct before you upgrade.
  • table column names changed. A table field that maps to a differently named column now comes back under the name you asked for. For example, table time returns a time column, not timestamp. Check saved searches, dashboards and integrations that read the old name. A command after table that needs timestamp won't find it unless the table keeps it.
  • Saved searches. Searches saved before v2.0.18 stored only their resolved window, so they reopen on that fixed past window. To make one relative again, open it, choose a preset and save it again.
  • Audit search is server-side. The search box and q on GET /api/audit and /api/audit/export match text across message, action, user, IP and details. Results can include rows that weren't on screen before.

On this page