Hi team,
Since the release of Conan Exiles: Enhanced, I’ve been seeing serious stability issues with the dedicated server binaries. This appears to affect both the native Linux dedicated server and the Windows dedicated server running via Wine.
The main issue is that the server’s memory usage is now significantly higher than before, and the process eventually gets killed due to running out of memory.
Before the Enhanced update, my dedicated server could run comfortably with an 8 GB memory allocation. With around 5 players online, typical usage was approximately 5–6 GB, sometimes pushing toward 7 GB, but it did not OOM.
Since the Enhanced update, the server appears to require closer to 9 GB+ as a practical minimum, with usage remaining consistently high. In my case, it has caused repeated out-of-memory kills on the host. This seems excessive even accounting for the UE5 migration.
My current setup:
Game: Conan Exiles: Enhanced
Server binary tested: Native Linux dedicated server
Also observed with: Windows dedicated server via Wine
Panel: Pterodactyl
Container runtime: Docker
Host virtualization: Proxmox
Guest OS: Ubuntu VM
Hardware: HP ProLiant ML310e Gen8 v2
Players: Approximately 5 active players during typical use
The server is running inside a Pterodactyl-managed Docker container. Docker/Pterodactyl overhead is minimal and does not appear to be the primary contributor. The issue appears to be the Conan server process itself consuming substantially more memory than it did before the Enhanced update.
Example kernel log entries from the Ubuntu VM show the Conan server being killed by the Linux OOM killer:
Out of memory: Killed process (ConanSandboxSer)
Out of memory: Killed process (GameThread)
oom-kill: task=ConanSandboxSer
In one captured case, the server process was killed with memory usage around:
ConanSandboxSer total-vm: 13750008kB
anon-rss: 6156044kB
Other OOM events showed similar behaviour, with the Conan server repeatedly being selected as the process to kill. When this happens, Pterodactyl/Wings detects the server as crashed and restarts it, leading to a crash/restart loop if the memory pressure persists.
I have since increased the VM and container memory limits to reduce the immediate OOMs, but the important point is that the required memory headroom has increased dramatically compared with the pre-Enhanced dedicated server.
Current mitigated configuration:
VM memory: ~15 GB assigned
Container memory limit: ~12.6 GiB
Swap: 4 GB
Proxmox ballooning: disabled
With this configuration the server is more stable, but the memory requirement is much higher than before. This is especially problematic for smaller private servers that previously ran fine with 8 GB.
Summary of issue:
Before Enhanced:
- 8 GB allocation was sufficient
- 5 players used roughly 5–6 GB
- Occasional peaks around 7 GB
- No regular OOM crashes
After Enhanced:
- Server appears to need roughly 9 GB+ minimum
- Memory usage remains consistently high
- 8 GB allocation leads to OOM kills
- Crash/restart loops can occur under Pterodactyl/Wings
- Reproduced/observed with both Linux and Windows-via-Wine dedicated server setups
Could the team please investigate whether there is a memory regression or leak in the Enhanced dedicated server binaries? The increase in baseline memory usage seems substantial, and it is causing real stability problems for community-hosted dedicated servers.
I’m happy to provide additional logs, Pterodactyl/Wings output, or kernel OOM traces if useful.