-
Notifications
You must be signed in to change notification settings - Fork 7
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
gpu: #1923's /dev/kfd guard over-fires on non-llama Vulkan runtimes — kokoro/whisper.cpp/ComfyUI slots refused on kfd-less AMD boxes
needs-triageMaintainer needs to evaluateMaintainer needs to evaluateStatus: Open.#1941 In Hal0ai/hal0;updater: failure-path stderr keeps only a 4000-char tail — a post-cosign failure with a long cleanup tail loses the verification breadcrumbs
needs-triageMaintainer needs to evaluateMaintainer needs to evaluateStatus: Open.#1940 In Hal0ai/hal0;api: image_status has no 'unknown' member — an unanswerable podman-seam read is indistinguishable from a genuinely missing image
needs-triageMaintainer needs to evaluateMaintainer needs to evaluateStatus: Open.#1939 In Hal0ai/hal0;install: four more hal0-api write paths have no perms.py ownership rows — images/cache, dashboard-layout.json, update-jobs, chat_templates
needs-triageMaintainer needs to evaluateMaintainer needs to evaluateStatus: Open.#1938 In Hal0ai/hal0;runner ade07ba: segfault (SIGSEGV) at model load when the container has zero GPU devices mapped — crashes before HTTP binds
needs-triageMaintainer needs to evaluateMaintainer needs to evaluateStatus: Open.#1936 In Hal0ai/hal0;agents: anchor_window ceiling diagnosis misses id-keyed slot TOMLs (read_slot_ceiling assumes <display-name>.toml)
needs-triageMaintainer needs to evaluateMaintainer needs to evaluateStatus: Open.#1932 In Hal0ai/hal0;memory: extraction preflight trusts wrapper.extraction_slot (configured intent), not the live hindsight-api drop-in
needs-triageMaintainer needs to evaluateMaintainer needs to evaluateStatus: Open.#1931 In Hal0ai/hal0;memory: MCP memory_add path has no extraction-ctx preflight — #1903 defect is identical on /mcp/memory
needs-triageMaintainer needs to evaluateMaintainer needs to evaluateStatus: Open.#1930 In Hal0ai/hal0;ui: memory-map GTT-cap fallback reads dead keys (rawHw.unified_memory_mb, stats gtt_total_mb) — pool silently becomes system RAM while labelled "GPU pool (GTT)"
needs-triageMaintainer needs to evaluateMaintainer needs to evaluateStatus: Open.#1929 In Hal0ai/hal0;ui: memory ruler header and bar derive "free" from different bases (model sum vs raw GTT) — same-screen self-contradiction survives #1900
needs-triageMaintainer needs to evaluateMaintainer needs to evaluateStatus: Open.#1928 In Hal0ai/hal0;ci: python (3.12) job has no timeout-minutes — an apt stall hangs it toward GitHub's 6-hour default
needs-triageMaintainer needs to evaluateMaintainer needs to evaluateStatus: Open.#1927 In Hal0ai/hal0;ci: γ-suite timeout-minutes:12 has no real headroom — green runs hit 10-11.3 min and slow runners randomly block PRs
needs-triageMaintainer needs to evaluateMaintainer needs to evaluateStatus: Open.#1926 In Hal0ai/hal0;