Environment
- Installation method: Official Greenbone Community Containers
- Host: Ubuntu
- gvmd: 26.35.0 (Manager DB revision 280)
- ospd-openvas: 22.10.4
- Image registry: registry.community.greenbone.net/community/*
Problem
Scans complete successfully on the scanner side and reach status Done, but immediately afterward gvmd aborts and the task is changed to Interrupted.
Relevant gvmd log excerpt:
event task:MESSAGE:... Status of task ... has changed to Processing
event task:MESSAGE:... Status of task ... has changed to Done
event alert:MESSAGE:... The alert nvas CSV was triggered (Event: Task status changed to 'Done', Condition: Always)
md manage:WARNING:... sql_exec_internal: PQexec failed: ERROR: column "severity_error" does not exist
LINE 1: ...t) FROM results WHERE report = 45 AND severity != SEVERITY_E...
^
(7)
md manage:WARNING:... sql_exec_internal: SQL: SELECT count (DISTINCT nvt) FROM results WHERE report = 45 AND severity != SEVERITY_ERROR;
md manage:WARNING:... sql_x: sql_exec_internal failed
... backtrace ...
md manage:MESSAGE:... Received Aborted signal
md manage:WARNING:... manage_cleanup_process_error: Error exit, setting running task to Interrupted
ospd-openvas logs show the scan finishing normally (“Host scan finished” / “Scan finished”). No blocks or interference from CrowdStrike Falcon (nothing in the Falcon console).
Analysis
The scan itself finishes correctly. The failure occurs in post-processing (result counting / alert path). The SQL sent to PostgreSQL contains the literal string SEVERITY_ERROR instead of the expanded numeric constant. This causes the query to fail, gvmd aborts, and the task is forced to Interrupted.
Questions
- Is this a known issue in gvmd 26.35.0 / DB revision 280?
- Is there a workaround (other than disabling the alert that triggers on Done)?
- Should a GitHub issue be opened against gvmd?
Happy to provide additional logs or test any suggested fixes.
