System Information
| Field | Value |
|---|---|
| Operating System | Linux - Debian GNU/Linux 13 on x86_64 |
| Product | AMP ‘Proteus’ v2.8.0.4 (Mainline) |
| Virtualization | Docker |
| Application | Dune Awakening |
| Module | GenericModule |
| Running in Container | Yes |
| Current State | Ready |
Problem Description
Issue
Fresh Dune Awakening instance won’t fully start. Backend comes up fine (postgres, both rabbitmq brokers, text-router, mock-k8s, director, gateway all report OK), but all three game maps fail:
ue5-Survival_1: no PID file — service did not start
ue5-Overmap: no PID file — service did not start
ue5-DeepDesert_1: no PID file — service did not start
UE5 game port pool stays at 0/30 on the status page. Everything else listens.
The first-run schema load is throwing an AssertionError and bailing before it builds the base schema. From the prestart section of the console:
[prestart] [INFO] invoking resetdb (loads schema via ToolsDB)…
Traceback (most recent call last):
…
File “/AMP/duneawakening/extracted/db-utils/root/PSQL/ToolsDB/setupdb.py”, line 56, in _import_module
assert os.path.exists(module_path)
AssertionError
[prestart] [INFO] Database schema loaded: success.
[prestart] [INFO] Seeding world partitions…
ERROR: relation “dune.world_partition” does not exist
Note that prestart prints “Database schema loaded: success” right after the traceback, so the failure is getting swallowed and nothing flags it until the maps don’t come up.
I dug into it. setupdb.py line 55-56 checks for an init.py under settings.module_path. settings.py line 86 sets:
module_path = self.args.schema_path or (config.paths.project / ‘Database’)
The schema package it wants is actually present on disk here:
/AMP/duneawakening/extracted/db-utils/root/DuneSandbox/Database/init.py
So the file exists, but the first-run resetdb call is resolving module_path to a Database dir that doesn’t have it, and the assert fails. Looks like schema_path isn’t being passed, or config.paths.project is resolving to the wrong root, on that initial call.
the base partition seed never runs, so world_partition only ends up with the 3 DLC rows the migrate step force-inserts:
select partition_id, map from dune.world_partition order by partition_id;
partition_id | map
--------------±--------------------------------
29 | CB_Overland_S_08
30 | CB_Dungeon_ThePit
31 | CB_SurvivalChallenge_Station_15
Partitions 1-28 (Survival_1, Overmap, DeepDesert_1, etc.) are never created, so those maps have nothing to boot against and never write a PID.
What I’ve tried:
Deleted and recreated the instance from scratch
Full clean re-extract (wiped extracted/, re-ran Update)
Full state wipe (removed server/state so first-run creates the dune role and DB from nothing)
Hardware isn’t the issue: AVX2 present, plenty of RAM and disk, all backend services healthy.
Happy to pull any additional logs or run whatever you need.
Reproduction Steps
- Create Instance
- Manage Instance
- Configure Token
- Run Updates
- Stop and Start Instance
- Start Server
@Mike Any insights you could share would be amazing.