llm-inference: size ctx-size/parallel for aux task offload

Previous ctx-size=8192/parallel=4 gave 2048 tokens/slot, too small for
context compression inputs (observed live rejection at 3826 tokens).

Measured VRAM on astro-orbiter (RTX 3090 24GB): weights ~17GB resident,
~294KiB/token pool-wide for KV cache+buffers at prior sizing.

New: ctx-size=16384, parallel=2 -> 8192 tokens/slot (matches model's
native n_ctx_train max). Projected VRAM ~21.8GB, ~2.7GB headroom.

Applied directly via ansible-playbook (Semaphore currently broken --
fix tracked separately).

ansible/roles/llm-inference/defaults/main.yml

@@ -21,9 +21,28 @@ llm_hf_model: google/gemma-2-27b-it
 # llama-server serve
 llm_serve_port: 8000
 llm_serve_host: "0.0.0.0"
-llm_max_model_len: 8192
+# NOTE (2026-08-05): --ctx-size is llama.cpp's TOTAL KV cache pool, divided
+# evenly across --parallel slots (per-slot context = ctx-size / parallel).
+# Previous 8192/4=2048 tokens-per-slot was too small for aux task offload
+# (context compression) and caused live rejections: "request (3826 tokens)
+# exceeds the available context size (2048 tokens)".
+# Sized against measured VRAM on astro-orbiter (RTX 3090, 24576MiB total):
+#   - Weights (Q4_K_M, 27B) ~16998MiB resident.
+#   - At ctx-size=8192/parallel=4, total llama-server VRAM = 19404MiB
+#     (nvidia-smi), i.e. ~2406MiB for KV cache + compute buffers at 8192
+#     total context tokens -> ~294KiB/token (pool-wide, incl. buffers).
+#   - Model n_ctx_train=8192 is the native max; per-slot context beyond
+#     this degrades coherence, so per-slot should cap at 8192.
+#   - New sizing: ctx-size=16384, parallel=2 -> 8192 tokens/slot (native
+#     max, covers compression's multi-thousand-token inputs with margin).
+#     Projected VRAM: 16998 + (~294KiB/token * 16384) ≈ 21.8GB used,
+#     leaving ~2.7GB headroom on the 24GB card.
+#   - parallel=2 (not 4) trades some concurrency for correct per-slot
+#     context; 2 concurrent aux-task requests is enough headroom before
+#     the known "3+ simultaneous compressions" GPU bottleneck kicks in.
+llm_max_model_len: 16384
 llm_gpu_layers: 99              # offload all layers to GPU
-llm_parallel_slots: 4           # concurrent request slots
+llm_parallel_slots: 2           # concurrent request slots -> 8192 tokens/slot (ctx-size / parallel)
 llm_gguf_dir: /home/jarvis/models
 llm_gguf_path: /home/jarvis/models/gemma-2-27b-it-Q4_K_M.gguf