Describe the bug

OrbStack Helper terminates unexpectedly with SIGKILL while Docker development/test workloads are active.

The most recent occurrence was at 2026-07-19 14:20:39 +08:00:

Host exit diagnostics:
- reason: killed (SIGKILL)
- raw wait status: 9 (0x00000009)
- raw flags: 0x86000000
- exit detail: none

This has happened more than once. After the previous occurrence, restarting OrbStack left the Docker images and data image present, but all previous container records were missing from docker ps -a. The images consequently appeared as "Unused" in the OrbStack UI.

The existing OrbStack data image was mounted again with the same Btrfs filesystem UUID, so this did not look like a complete data-image reset. After recovery, the log also contained errors for stale/missing Docker volume directories, for example:

failed to mount docker volume: open volume dir '/volumes/<volume-id>/_data': no such file or directory

A persistent Docker container should not lose its metadata merely because the OrbStack host process was killed.

To Reproduce

The failure is intermittent and I do not yet have a fully deterministic reproducer.

1. Start OrbStack normally.
2. Run active Docker development workloads, including Docker Compose services (PostgreSQL, Redis, and MinIO) and test scripts that create/remove PostgreSQL containers.
3. Continue development/test activity for several minutes.
4. OrbStack displays: "Stopped unexpectedly: killed (SIGKILL)".
5. Restart OrbStack.
6. Check docker ps -a and the OrbStack Containers/Images/Volumes pages.
7. On the previous occurrence, the earlier containers were missing while their images remained and were listed as "Unused".

Expected behavior

OrbStack Helper should remain running during normal Docker development workloads.

If the host process is terminated unexpectedly, Docker should recover consistently after restart: persistent containers should remain visible in docker ps -a (normally in an exited state), and existing volume metadata should continue to reference valid volume data directories.

Diagnostic report (REQUIRED)

OrbStack info:
Version: 2.2.1
Commit: 0e182b501fcd9e05b99ffb363fce03610390c400 (v2.2.1)

System info:
macOS: 26.5.2 (25F84)
CPU: arm64, 10 cores
CPU model: Apple M4
Model: Mac16,10
Memory: 16 GiB

Full report: https://orbstack.dev/_admin/diag/orbstack-diagreport_2026-07-19T06-25-23.692864Z.zip

Screenshots and additional context (optional)

Additional observations:

• The most recent OrbStack Helper PID was 23511 and it exited at 14:20:39 with SIGKILL.
• OrbStack's own vmgr log ends without an internal panic or handled shutdown.
• macOS unified logs around the same time contain many JETSAM_REASON_MEMORY_IDLE_EXIT events. However, no new JetsamEvent report was generated specifically for OrbStack Helper, so I cannot confirm which process or subsystem sent SIGKILL.
• At the time of investigation, the host still had about 63 GiB of disk space available.
• The port-forwarding errors shown near the end of vmgr.log appear to be secondary to the VM/network restart rather than the initiating failure.
• This issue is particularly damaging because the previous occurrence was followed by loss of all Docker container metadata, while images remained.

The uploaded diagnostic report contains the complete logs.