What changed in v0.9.4
The theme of this release is certainty: AASM now proves what is running before it touches anything, and refuses with a reason when it cannot. Around that core: chat that crosses machines, Linux crash recovery that works for the first time, and licensing that never takes a running fleet down.
Four-state liveness — refuses instead of guessing
“Is this server running?” now has four honest answers: yes, no, ambiguous, and I could not check. Before, “I could not check” quietly became “not running” — which is how a manager writes STOPPED over a live server, or starts a second copy on the same save. Every destructive action — start, stop, delete, update, rollback, restore, move — now asks first, and if the install’s process state cannot be proven it refuses and tells you why. A failed process scan can no longer be mistaken for an empty one.
What you will see: refusal messages that are new to you, such as “the install’s process state could not be verified — refusing to spawn blind.” That is the safety work doing its job, not a fault. Retry shortly; if it persists, the process-enumeration tools on that machine are failing and that is the thing to investigate.
Cross-machine chat
Cross-chat no longer stops at the edge of one PC. One AASM instance acts as the hub, the others link to it, and chat flows between every server in the fleet regardless of machine. Messages carry their origin so nothing loops or echoes; links are authenticated and fingerprint-confirmed; two machines both claiming to be the hub is detected and surfaced rather than silently double-posting. Chat never queues across an outage — lines sent while a link is down are dropped, not replayed ten minutes late.
Linux: auto-restart on crash, working for the first time
It had never actually fired on Linux. A dead server was indistinguishable from an unreadable one, so AASM retried forever while the interface showed it running — we watched it retry 33 times against a server that had been dead for minutes. Fixed at the root: a missing process now counts as proof of exit, anything genuinely unreadable still means don’t guess, and a dead server is detected in about 35 seconds and restarted — verified end to end on a real Proton server. Linux also now detects and adopts servers you started outside AASM, matching Windows.
Licence outages no longer take you down
Two operators hit this: our licensing host had DNS trouble and AASM shut itself down thirty seconds later, with servers running. It cannot now. AASM carries a 6-hour offline grace window — it keeps working, retries hourly, and shows how long is left — and it will not close while your servers are running. If grace runs out mid-session it stays open and keeps monitoring, blocking only new server starts and install changes until it can check in again. Your ARK servers were always safe (they run independently); what you were losing was monitoring, scheduled restarts, backups, crash recovery and RCON.
More requested capabilities
Discord bot: tribe log, admin log and game log can each go to their own channel (default off), with proper classification and first-poll priming. Network and disk I/O in the dashboard alongside CPU and memory, on both platforms. Exact server-build selection and pinning with direction-aware warnings. Countdown intervals on a sensible schedule plus a 3/2/1 finish across bulk operations. Player-profile backups copyable between servers. MOTD update-checker quotes that no longer break your config.
Reliability and community-reported fixes
- Scheduled restarts no longer leave servers offline: a maintenance restart could mark a still-running server STOPPED and never bring it back. Maintenance now waits for a confirmed transition and aborts on ambiguity.
- Per-map mod lists are no longer overwritten fleet-wide: mods are now a separate, default-off sync category — your per-map lists stay yours unless you explicitly opt in.
- Re-adoption after a restart works for installs registered under Windows short paths (
PROGRA~1style), junction-safe. - The T-10 minute maintenance warning now fires when the timing lands exactly on a boundary.
- Version checks now report when they last succeeded and colour the result honestly, instead of implying a check that never ran.
- Licence-server DNS failures no longer shut AASM down with servers running (see above).
- Companion app: the bundled ASA RCON Manager is version-synced to v0.9.4.
Platform and file details
| Platform | Package | Size | Requirements |
|---|---|---|---|
| Windows | .zip | 377.5 MB | Windows 10/11 or Windows Server 2019+ |
| Linux | .tgz · AppImage · .deb | 290.8 MB | 64-bit graphical session; Ubuntu 22.04+ or Debian 12+ recommended |
For either platform, use at least 16 GB RAM. AASM recommends 32 GB when running multiple ASA servers, plus SSD storage and sufficient free space for server files, staging data and backups.
Upgrade guidance
- Stop active server maintenance and wait for backups or updates already in progress to finish.
- Create a current backup of every world and any shared cluster data.
- Update in-app through Settings → Check for Updates, or download the correct platform archive from this domain and verify its SHA-256 fingerprint.
- Confirm v0.9.4 in AASM before restarting the fleet.
- Expect refusal messages you have not seen before (“process state could not be verified”). That is the new safety work refusing to act blind; retry shortly rather than forcing anything.