update CAR/VAR vars and others - #315
Conversation
* chore: inc version * rm: cerviguard image processor plugin * chore: inc version
* feat: integration of both vllm and llama_cpp for inference * chore: inc ver
* fix: use postponed request for local serving API * fix: endpoints cleanup * fix: add analysis data in response * fix: add cerviguard to ai engines * fix: update ai_engine prop in config * fix: update alias * fix: rm alias * chore: inc version
* fix: container autostart policy * fix: shell injection vulnerability & remove deprecated flag * fix: add exponential backoff delay for tunnels * fix: autoupdate * feat: persist container stop state between plugin restarts * fix: paused log interval * fix: add numpy style docstrings * chore: inc version * fix: docstrings
* fix: move memory formatter functions to utils * fix: add memory and cpu limits for container app runner * fix: container start log * fix: rename config variable * fix: make distance between methods to 2 new lines * fix: add gpu support to CAR * chore: increment version * fix: container start log * fix: gpu logs cleanup * fix: docstring * fix: logs color cleanup * fix: scale mem reservation
chore: prep for mainnet
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| container_resources = self.cfg_container_resources | ||
| if isinstance(container_resources, dict) and len(container_resources) > 0: | ||
| self._cpu_limit = container_resources.get("cpu", DEFAULT_CPU_LIMIT) | ||
| self._cpu_limit = int(container_resources.get("cpu", DEFAULT_CPU_LIMIT)) |
There was a problem hiding this comment.
In _setup_resource_limits_and_ports the CPU limit is now coerced with int(container_resources.get("cpu"...)), truncating fractional values that the config explicitly supports (e.g., "0.5" for half a core). With this cast a 0.5 setting becomes 0 and nano_cpu_limit later passed to docker_client.containers.run is 0, so requested throttling is ignored or the container fails to start when users specify fractional CPUs. Keeping the original float/string value (or parsing to float) avoids silently dropping CPU limits.
Useful? React with 👍 / 👎.
* feat: support for sqlcoder and llm inference api * fix: removed date_string from prompt * fix: naming of endpoints * fix: ver change
4 participants
No description provided.