PUBLIC/gpu-turnstile
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 3 Wiki Activity
9 Commits 1 Branch 9 Tags
20d5439b5f0998b81baced058751d60e2b6d55e0
Commit Graph
4 Commits
Author SHA1 Message Date
mram 20d5439b5f Retry upstream connection-refused with exponential backoff 
A refused dial (service down/restarting) is retried with a wait that
doubles from BACKOFF_INITIAL (1s) up to BACKOFF_MAX (60s) until the
upstream answers or the client disconnects. Handles the Windows WSA
errno (10061) as well as POSIX ECONNREFUSED. compose.yaml.example now
uses host.docker.internal like the working local deployment.
 2026-09-20 19:45:48 +02:00
mram cf9be55a6c Make poll intervals and operational timeouts configurable
ci / test (push) Successful in 47s
Details
ci / docker (push) Failing after 30s
Details 
New env vars: UNLOAD_POLL_INTERVAL, HISTORY_POLL_INTERVAL, PROBE_TIMEOUT,
FREE_TIMEOUT, WARM_TIMEOUT, SHUTDOWN_TIMEOUT, PROMPT_CAPTURE_LIMIT.
Defaults unchanged; invalid values fail fast at startup.
 2026-09-20 19:21:25 +02:00
mram db11da8307 Rename to gpu-turnstile; swap port roles 
The proxy now listens on the standard service ports (Ollama :11434,
ComfyUI :8188) and the actual services move one port up (:11435, :8189).
Metric prefix is now gpu_turnstile_.
 2026-09-20 18:10:22 +02:00
mram 065c294a96 gpulock: GPU arbitration proxy for Ollama + ComfyUI 
Implements SPEC.md: two listeners, one writer-preferring two-mode lock,
Ollama unload before image jobs, ComfyUI history polling + VRAM free,
optional model warm-up, healthz/metrics endpoints, streaming-safe
reverse proxies, Dockerfile and Gitea Actions CI.
 2026-09-20 18:05:20 +02:00