I see several cases of this error happening of users of AMP in docker for 8-10+ instances, my Claude Opus managed to find a solution, I hope this helps someone else as well. CREATING THIS ISSUE FOR VISIBILITY.
Symptom
Creating a new instance fails with:
AMP configuration failed with return code 135
The new instance's log ends at Performing activation, please wait... with no exception data. Existing instances keep working fine.
Root cause
Exit code 135 = SIGBUS (128 + 7). AMP is .NET and uses /dev/shm/dotnet_* for memory-mapped IPC. Docker defaults /dev/shm to 64 MB, which fills up once several AMP instances are running. The activation step needs more shm, mmap pages can't be backed, SIGBUS kills AMP.
Verify inside the container: df -h /dev/shm — if it's near full, this is your bug.
Fix
Add to docker-compose.yml:
services:
amp:
shm_size: '1gb'Or docker run --shm-size=1g .... Then recreate the container (docker compose down && up -d); a plain restart doesn't resize tmpfs.
Symptom
Creating a new instance fails with:
The new instance's log ends at
Performing activation, please wait...with no exception data. Existing instances keep working fine.Root cause
Exit code 135 = SIGBUS (128 + 7). AMP is .NET and uses
/dev/shm/dotnet_*for memory-mapped IPC. Docker defaults/dev/shmto 64 MB, which fills up once several AMP instances are running. The activation step needs more shm, mmap pages can't be backed, SIGBUS kills AMP.Verify inside the container:
df -h /dev/shm— if it's near full, this is your bug.Fix
Add to
docker-compose.yml:Or
docker run --shm-size=1g .... Then recreate the container (docker compose down && up -d); a plain restart doesn't resize tmpfs.