Our community dedicated server (Namalsk, 1.30.164078, CFTools Architect, Windows Server 2025 / 26100) hard-crashes repeatedly with an identical access violation. 5 crashes on 19 Oct 2026, all at the same instruction:
Access violation. Illegal read by DayZServer_x64+0x4dfa93 at 0x10
00007ff7`8719fa93 488b4810 mov rcx, qword ptr [rax+10h] (rax = 0)
Failure bucket: INVALID_POINTER_READ_c0000005_DayZServer_x64.exe!Unknown
Fault code bytes: 48 8B 48 10 48 85 C9 74 08 E8 0F E7 D4 FF 48 8B
Crash times (UK time, 19 Oct): 14:52, 17:48, 18:45, 19:35, 20:05. The call stack is identical each time (same top 6 frames, main thread).
We captured two FULL memory dumps with ProcDump (-ma -e 1 -f C0000005). Reading the RTTI of the objects on the crashing stack, both crashes happen inside a Bullet ray test against the terrain heightfield, issued by an AI creature's character controller:
Dump 1 (19:34:58): HeightfieldTerrainShape, btCollisionWorld::rayTestSingle (LocalInfoAdder2), btSingleRayCallback, enf::RayExceptOne, enf::MyDiscreteDynamicsWorld, enf::BulletCharacterCtrlSimple, DayZInfected (infected near ~6870, 11290)
Dump 2 (20:05:45): HeightfieldTerrainShape, btSingleRayCallback, enf::RayExceptOne, enf::MyDiscreteDynamicsWorld, enf::BulletCharacterCtrlSimple, DayZAnimal + BehaviourGroupPredators (wolf near ~8275, 10370)
In both dumps rsi = the HeightfieldTerrainShape and rax (the pointer being read) is null, so it looks like a terrain heightfield/cell lookup returning null during a creature ground/collision ray test.
No mod script is on the crashing stack. This crash did not occur before 1.30. Namalsk's 1.30 update uses the new terrain-holes feature (Athena-3, Athena-2, Athena Research Institute), which may be relevant, but neither crash location is at those areas.
Several crashes occurred within ~1-3 s of a player connecting or being kicked by BattlEye ("Client not responding"), but many such events do not crash, so this may be coincidental. One crash (14:52) happened ~1 minute after server start.