From ef324418207842ed175b6a7b98065f05bf0ce8d3 Mon Sep 17 00:00:00 2001 From: dvcdsys Date: Mon, 22 Jun 2026 21:59:29 +0100 Subject: [PATCH] chore(server/cuda): bump pinned llama.cpp digest to 2ea4be99 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Upstream ghcr.io/ggml-org/llama.cpp:server-cuda moved past the pinned digest. Bump 67bf4625 → 2ea4be99 (current latest, verified via imagetools inspect). The digest is pinned because the image layout has shifted before (the libllama-server-impl.so split that took prod down with exit 127) — so this must be validated with a scout-cuda rebuild + a GPU smoke test that llama-server resolves all libs and embeddings actually run on the GPU before it goes anywhere near a release. Addresses the weekly llama.cpp-pin-freshness issues #84 #83 #82 #58. Co-Authored-By: Claude Opus 4.8 --- server/Dockerfile.cuda | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/server/Dockerfile.cuda b/server/Dockerfile.cuda index d1339198..5dc51078 100644 --- a/server/Dockerfile.cuda +++ b/server/Dockerfile.cuda @@ -91,7 +91,7 @@ RUN curl -fsSL \ # missing libllama-server-impl.so). Bump this digest deliberately and re-run # `make scout-cuda` to validate. Resolve a new digest with: # docker buildx imagetools inspect ghcr.io/ggml-org/llama.cpp:server-cuda -FROM ghcr.io/ggml-org/llama.cpp:server-cuda@sha256:67bf4625a2f788864cecfabb30c33f4cee97f055326e008e1d608c2ba44a1c94 AS llama-source +FROM ghcr.io/ggml-org/llama.cpp:server-cuda@sha256:2ea4be99dab097e84527ec552b8e5a552095cee3625b2cce8c052eb816a6c941 AS llama-source # ── Stage 3: extract CUDA shared libraries ───────────────────────────────── # Install the CUDA libs here, then COPY individual .so files into the