LCORE-431: how to build the lightspeed core image #391
Conversation
WalkthroughUpdated docs and container configurations: added "Building Container Images" docs describing Server vs Library deployment modes, adjusted container images (added C/C++ tools in test image), changed service bind/address and llama_stack URL, and added a new run.yaml with sqlite-backed stores and provider configurations. Changes
Sequence Diagram(s)sequenceDiagram
participant Dev as Developer
participant Builder as Builder Stage
participant Final as Final Image
Dev->>Builder: define Containerfile / test.containerfile
Builder->>Builder: install runtime + build tools (gcc, make)
Builder-->>Final: copy runtime artifacts (build-only tools excluded)
sequenceDiagram
participant Client
participant Core as lightspeed-core
participant Stack as llama-stack
participant OpenAI
rect rgb(240,240,255)
note over Core,Stack: Server Mode
Client->>Core: HTTP request
Core->>Stack: HTTP call (http://llama-stack:8321)
Stack->>OpenAI: Provider inference request
OpenAI-->>Stack: Response
Stack-->>Core: Result
Core-->>Client: Response
end
rect rgb(240,255,240)
note over Core: Library Mode
Client->>Core: HTTP request
Core->>OpenAI: Inference via embedded llama-stack library
OpenAI-->>Core: Response
Core-->>Client: Response
end
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related PRs
Suggested reviewers
Poem
Tip 🔌 Remote MCP (Model Context Protocol) integration is now available!Pro plan users can now connect to remote MCP servers from the Integrations page. Connect with popular remote MCPs such as Notion and Linear to add more context to your reviews and chats. ✨ Finishing Touches🧪 Generate unit tests
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. CodeRabbit Commands (Invoked using PR/Issue comments)Type Other keywords and placeholders
CodeRabbit Configuration File (
|
gallettilance
changed the title
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 7
🔭 Outside diff range comments (2)
test.containerfile (1)
27-43: Out-of-lock uv pip installs undermine reproducibilityInstalling a large set of packages with uv pip install after uv sync --locked bypasses uv.lock, leading to non-reproducible images and potential supply-chain drift. Prefer declaring these dependencies in pyproject.toml and re-locking so uv sync --locked can reproduce the environment deterministically. If you need optional tooling (e.g., test-only), consider extras or separate lock files.
I can generate a PR to move these to pyproject and refresh uv.lock. Want me to proceed?
lightspeed-stack.yaml (1)
3-7: Secure service exposure risk in lightspeed-stack.yamlBinding the service to all interfaces with authentication disabled allows unrestricted access:
File: lightspeed-stack.yaml
Lines: 3–7Current config:
host: 0.0.0.0 port: 8080 auth_enabled: false workers: 1 color_log: trueRecommendations:
- If the service must remain internal only, restrict binding:
host: 127.0.0.1- Otherwise, enable basic auth:
auth_enabled: true auth: username: myuser password: ${LS_PASSWORD}- Document running behind a reverse proxy (e.g., Nginx) with TLS termination and IP whitelisting.
- Add a minimal example in your README or docs to guide secure deployment.
🧹 Nitpick comments (5)
test.containerfile (1)
22-23: Remove redundant uv -h layerThis layer only prints help and adds no value to the image; drop it to speed up builds.
-RUN uv -hREADME.md (2)
312-323: Prefer ‘docker compose’ over legacy ‘docker-compose’The plugin-based docker compose is the default going forward. Update the command for consistency with modern Docker installations.
-docker-compose up --build +docker compose up --build
355-356: Double-check the /v1/models endpoint existsIf the service doesn’t expose /v1/models, the verification step will fail. Use readiness/liveness as an alternative if models aren’t listed.
Would you like me to adjust the README to curl /v1/readiness instead?
run.yaml (1)
87-89: Avoid committing placeholder secrets in configEven though obfuscated, openai_api_key: '********' can cause confusion. Prefer referencing an environment variable consistently, or omit the field if unused.
- - config: - openai_api_key: '********' + - config: + openai_api_key: ${env.OPENAI_API_KEY}Containerfile (1)
60-66: PATH precedence ensures venv binaries; consider explicit python path for clarityENV PATH="/app-root/.venv/bin:$PATH" is correct. Optionally, set ENTRYPOINT to the venv’s python to avoid relying on PATH resolution.
-ENTRYPOINT ["python3.12", "src/lightspeed_stack.py"] +ENTRYPOINT ["/app-root/.venv/bin/python", "src/lightspeed_stack.py"]
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (5)
-
Containerfile(1 hunks) -
README.md(1 hunks) -
lightspeed-stack.yaml(2 hunks) -
run.yaml(1 hunks) -
test.containerfile(1 hunks)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: e2e_tests
🔇 Additional comments (3)
lightspeed-stack.yaml (1)
16-16: Ensure Compose service ‘llama-stack’ existsThere isn’t a Docker Compose file in this repo, but lightspeed-stack.yaml (line 16) points to
http://llama-stack:8321. Please verify that your external Compose setup defines a service or network alias named llama-stack, or update this URL to match your actual service name.• lightspeed-stack.yaml:16 –
url: http://llama-stack:8321run.yaml (2)
63-67: Environment variable expansion for OpenAI key: confirm supported syntaxConfirm that ${env.OPENAI_API_KEY} is the correct interpolation syntax expected by your llama-stack version. Some configs prefer ${env:OPENAI_API_KEY}.
I can search for examples in this repo or upstream configs to validate and adjust accordingly. Do you want a script to scan for similar patterns?
20-26: Relative DB paths under .llama directory: ensure write permissions and directory creationThese SQLite paths are relative (e.g., .llama/distributions/ollama/...). Make sure the working directory is writeable by the runtime user (1001) and that parent directories are created on startup or via entrypoint logic.
I can add a tiny init script to create these directories at container start if you’d like.
Also applies to: 31-37, 42-45, 49-52, 57-61, 93-96, 103-107
Containerfile
Outdated
| RUN microdnf install -y --nodocs --setopt=keepcache=0 --setopt=tsflags=nodocs \ | ||
| python3.12 python3.12-devel python3.12-pip git tar \ | ||
| gcc gcc-c++ make | ||
|
|
There was a problem hiding this comment.
🛠️ Refactor suggestion
Install curl and CA bundles; clean cached metadata
curl is used later (Line 24), and HTTPS requires CA bundles. Also, clean DNF metadata to reduce image size.
-RUN microdnf install -y --nodocs --setopt=keepcache=0 --setopt=tsflags=nodocs \
- python3.12 python3.12-devel python3.12-pip git tar \
- gcc gcc-c++ make
+RUN microdnf install -y --nodocs --setopt=keepcache=0 --setopt=tsflags=nodocs \
+ python3.12 python3.12-devel python3.12-pip git tar curl ca-certificates \
+ gcc gcc-c++ make && microdnf clean all🤖 Prompt for AI Agents
In Containerfile around lines 15 to 18, the image installs Python and build
tools but omits curl and CA bundles and does not clean DNF metadata; update the
RUN install line to include curl and an appropriate CA package (e.g.,
ca-certificates) so HTTPS calls work, and after installation run microdnf
cleanup to remove cached metadata (e.g., clean all/tmp metadata) to reduce image
size.
Containerfile
Outdated
| RUN curl -LsSf https://astral.sh/uv/install.sh | sh | ||
|
|
There was a problem hiding this comment.
🛠️ Refactor suggestion
Harden the uv install and pin the version
Same as test.containerfile: guard against partial downloads and pin uv to ensure reproducible builds.
-RUN curl -LsSf https://astral.sh/uv/install.sh | sh
+RUN set -euxo pipefail; curl -fsSL https://astral.sh/uv/install.sh | sh -s -- --version ${UV_VERSION:-latest}Add near top of this file:
ARG UV_VERSION=0.4.29🤖 Prompt for AI Agents
In Containerfile around lines 24-25, the RUN that pipes curl to sh needs to be
hardened and UV pinned: add an ARG UV_VERSION=0.4.29 near the top of the
Dockerfile, replace the direct pipe with a safe download that uses the pinned
UV_VERSION (e.g., curl -fSL to write to a temporary file), verify the download
(checksum or signature) before executing, and invoke the installer explicitly
from that trusted file rather than piping curl to sh to prevent partial-download
execution.
Containerfile
Outdated
| RUN uv sync --locked --no-dev && \ | ||
| uv pip install \ | ||
| opentelemetry-sdk \ | ||
| opentelemetry-exporter-otlp \ | ||
| opentelemetry-instrumentation \ | ||
| aiosqlite \ | ||
| litellm \ | ||
| blobfile \ | ||
| datasets \ | ||
| sqlalchemy \ | ||
| faiss-cpu \ | ||
| mcp \ | ||
| autoevals \ | ||
| psutil \ | ||
| torch \ | ||
| peft \ | ||
| trl |
There was a problem hiding this comment.
🛠️ Refactor suggestion
Out-of-lock installs reduce reproducibility; move deps into pyproject/uv.lock
uv pip install of many packages after uv sync --locked --no-dev side-steps lock guarantees. Prefer declaring these in pyproject.toml and re-locking so uv sync --locked fully reproduces the environment. This also makes dependency provenance auditable.
I can update pyproject.toml and uv.lock to include these packages and drop the uv pip install block. Want me to open a follow-up PR?
🤖 Prompt for AI Agents
In Containerfile around lines 26 to 42, the Dockerfile runs `uv pip install`
after `uv sync --locked --no-dev`, which bypasses the lockfile and undermines
reproducibility; instead add these packages to pyproject.toml (or the project's
uv configuration), run `uv lock` (or the project's equivalent) to update
uv.lock, commit the updated pyproject.toml and uv.lock, and remove the `uv pip
install` block from the Containerfile so `uv sync --locked --no-dev` alone
installs all declared dependencies from the lockfile.
| RUN curl -LsSf https://astral.sh/uv/install.sh | sh | ||
|
|
There was a problem hiding this comment.
🛠️ Refactor suggestion
Harden the curl | sh step and pin uv version
Piping curl to sh without pipefail can mask download failures; also pin uv for reproducible builds.
-RUN curl -LsSf https://astral.sh/uv/install.sh | sh
+RUN set -euxo pipefail; curl -fsSL https://astral.sh/uv/install.sh | sh -s -- --version ${UV_VERSION:-latest}Add this ARG near the top of the file (outside the selected range):
ARG UV_VERSION=0.4.29🤖 Prompt for AI Agents
In test.containerfile around lines 20-21, the RUN that pipes curl to sh is
unsafe and unpinned; add an ARG near the top of the file: UV_VERSION=0.4.29, and
update the RUN to enable strict failure handling and pass the pinned version to
the installer (use set -o pipefail -e and curl with --fail --show-error
--location, piping to sh -s -- "$UV_VERSION") so the build fails on download
errors and the uv version is reproducible.
90885f1 to
9fa1485
Compare
| @@ -1,25 +1,45 @@ | |||
| # vim: set filetype=dockerfile | |||
There was a problem hiding this comment.
Please drop changes in this file, I update mine PR to include this kind of changes. #292
| COPY uv.lock ./ | ||
| COPY LICENSE ./ | ||
| COPY README.md ./ | ||
| COPY src/ ./src/ |
There was a problem hiding this comment.
Same here: Please drop changes in this file, I update mine PR to include this kind of changes.
#292
README.md
Outdated
| **Build and run**: | ||
| ```bash | ||
| # Build lightspeed-core with embedded llama-stack | ||
| docker build -f Containerfile -t my-lightspeed-core:latest . |
There was a problem hiding this comment.
Please put here podman instead of docker (my opinion).
| -v ./run.yaml:/app-root/run.yaml:ro \ | ||
| -e OPENAI_API_KEY=your-api-key \ | ||
| my-lightspeed-core:latest | ||
| ``` |
There was a problem hiding this comment.
This command is more for macosx. Please also include this linux command:
podman run -p 8080:8080 -v ./lightspeed-stack.yaml:/app-root/lightspeed-stack.yaml:Z -v ./run.yaml:/app-root/run.yaml:Z -e OPENAI_API_KEY=$OPENAI_API_KEY my-lightspeed-core:latest
9fa1485 to
8d9ae4d
Compare
There was a problem hiding this comment.
Actionable comments posted: 0
♻️ Duplicate comments (1)
README.md (1)
576-582: Add a simple readiness wait before hitting /v1/modelsAvoids flaky failures if the service isn’t ready yet.
### Verify it's running properly A simple sanity check: ```bash +until curl -sf http://localhost:8080/v1/liveness > /dev/null; do + echo "Waiting for lightspeed-core to become ready..." + sleep 1 +done + curl -H "Accept: application/json" http://localhost:8080/v1/models</blockquote></details> </blockquote></details> <details> <summary>🧹 Nitpick comments (5)</summary><blockquote> <details> <summary>README.md (5)</summary><blockquote> `516-520`: **Unify naming and tighten wording (Llama Stack, Lightspeed Core; sentence clarity)** Minor editorial fixes: remove the hyphen in “Llama-Stack”, capitalize proper names, and clarify the sentence. ```diff -### Llama-Stack as Separate Service (Server Mode) +### Llama Stack as Separate Service (Server Mode) -When using llama-stack as a separate service, the existing `docker-compose.yaml` provides the complete setup. This builds two containers for lightspeed core and llama stack. +When using Llama Stack as a separate service, the existing `docker-compose.yaml` provides the complete setup. It builds two containers: Lightspeed Core and Llama Stack.
541-551: Fix minor wording and naming; use consistent “Llama Stack”Small grammar/style nits; also remove duplicate “directly”.
-### Llama-Stack as Library (Library Mode) +### Llama Stack as Library (Library Mode) -When embedding llama-stack directly in the container, use the existing `Containerfile` directly (this will not build the llama stack service in a separate container). First modify the `lightspeed-stack.yaml` config to use llama stack in library mode. +When embedding Llama Stack in the container, use the existing `Containerfile` (this will not build the Llama Stack service in a separate container). First modify the `lightspeed-stack.yaml` config to use library mode.
552-565: Linux SELinux note: add :Z label for podman volumes (keep RO mount)On Linux with SELinux, add the :Z label to avoid permission issues. You can combine :ro,Z.
podman run \ -p 8080:8080 \ - -v ./lightspeed-stack.yaml:/app-root/lightspeed-stack.yaml:ro \ - -v ./run.yaml:/app-root/run.yaml:ro \ + -v ./lightspeed-stack.yaml:/app-root/lightspeed-stack.yaml:ro,Z \ + -v ./run.yaml:/app-root/run.yaml:ro,Z \ -e OPENAI_API_KEY=your-api-key \ my-lightspeed-core:latest
566-574: Fix spelling (macOS) and remove SELinux label from macOS examplemacOS should be capitalized “macOS”. The :Z SELinux label isn’t needed on macOS; keep read-only mounts instead.
-For macosx users: +For macOS users: ```bash podman run \ -p 8080:8080 \ - -v ./lightspeed-stack.yaml:/app-root/lightspeed-stack.yaml:Z \ - -v ./run.yaml:/app-root/run.yaml:Z \ + -v ./lightspeed-stack.yaml:/app-root/lightspeed-stack.yaml:ro \ + -v ./run.yaml:/app-root/run.yaml:ro \ -e OPENAI_API_KEY=your-api-key \ my-lightspeed-core:latest--- `528-539`: **Ensure Compose filename consistency** The repository includes a single `docker-compose.yaml`, but the README’s example simply runs: ```bash podman compose up --buildWhile Podman Compose will pick up
docker-compose.yamlby default, it isn’t immediately obvious. To avoid confusion, consider one of the following:
- Explicitly reference the file in the command:
podman compose -f docker-compose.yaml up --build- Or rename
docker-compose.yamltocompose.yamland update the README accordingly.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
💡 Knowledge Base configuration:
- MCP integration is disabled by default for public repositories
- Jira integration is disabled by default for public repositories
- Linear integration is disabled by default for public repositories
You can enable these sources in your CodeRabbit configuration.
📒 Files selected for processing (5)
Containerfile(1 hunks)README.md(1 hunks)lightspeed-stack.yaml(2 hunks)run.yaml(1 hunks)test.containerfile(2 hunks)
✅ Files skipped from review due to trivial changes (1)
- Containerfile
🚧 Files skipped from review as they are similar to previous changes (3)
- lightspeed-stack.yaml
- test.containerfile
- run.yaml
🧰 Additional context used
🧠 Learnings (1)
📚 Learning: 2025-08-18T10:56:55.302Z
Learnt from: matysek
PR: lightspeed-core/lightspeed-stack#292
File: pyproject.toml:0-0
Timestamp: 2025-08-18T10:56:55.302Z
Learning: The lightspeed-stack project intentionally uses a "generic image" approach, bundling many dependencies directly in the base runtime image to work for everyone, rather than using lean base images with optional dependency groups.
Applied to files:
README.md
🪛 LanguageTool
README.md
[grammar] ~516-~516: There might be a mistake here.
Context: ...within lightspeed-core ### Llama-Stack as Separate Service (Server Mode) When us...
(QB_NEW_EN)
[grammar] ~566-~566: Ensure spelling is correct
Context: ... \ my-lightspeed-core:latest For macosx users:bash podman run \ -p 8080:8...
(QB_NEW_EN_ORTHOGRAPHY_ERROR_IDS_1)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (2)
- GitHub Check: build-pr
- GitHub Check: e2e_tests
🔇 Additional comments (2)
README.md (2)
509-515: Nice addition: clear section framing for container buildsThe new “Building Container Images” section neatly separates the two deployment modes. Good structure.
520-526: No documentation change needed:llama_stack.api_keyis supportedThe
api_keyfield underllama_stackis part of the configuration model (validated intests/unit/test_configuration.py) and is used when instantiatingAsyncLlamaStackClientinsrc/client.py. The example inREADME.mdis correct and should remain as is.
| -v ./lightspeed-stack.yaml:/app-root/lightspeed-stack.yaml:Z \ | ||
| -v ./run.yaml:/app-root/run.yaml:Z \ | ||
| -e OPENAI_API_KEY=your-api-key \ | ||
| my-lightspeed-core:latest |
There was a problem hiding this comment.
Just nitpick: This command is for linux. Please swap those commands
command macosx:
podman run -p 8080:8080 -v ./lightspeed-stack.yaml:/app-root/lightspeed-stack.yaml:ro \
-v ./run.yaml:/app-root/run.yaml:ro \
-e OPENAI_API_KEY=your-api-key \
my-lightspeed-core:latest
command for linux:
podman run -p 8080:8080 -v ./lightspeed-stack.yaml:/app-root/lightspeed-stack.yaml:Z -v ./run.yaml:/app-root/run.yaml:Z -e OPENAI_API_KEY=$OPENAI_API_KEY my-lightspeed-core:latest
gallettilance
force-pushed
the
docs-build-container
branch
from
8d9ae4d to
7ee8ae2
Compare
There was a problem hiding this comment.
Actionable comments posted: 0
♻️ Duplicate comments (1)
README.md (1)
552-564: Linux example: pass exported env var (per prior feedback)The line uses a literal placeholder; for Linux we should pass the env var value as suggested previously. This keeps it consistent with the earlier
export OPENAI_API_KEY=...step.- -e OPENAI_API_KEY=your-api-key \ + -e OPENAI_API_KEY=$OPENAI_API_KEY \
🧹 Nitpick comments (6)
README.md (6)
516-516: Naming consistency: prefer “Llama Stack” (no hyphen) in headingsElsewhere in the README you use “Llama Stack”. Align this header for consistency.
-### Llama-Stack as Separate Service (Server Mode) +### Llama Stack as Separate Service (Server Mode)
518-519: Wording and product naming tweaks
- “docker-compose.yaml” is fine, but the sentence reads a bit awkwardly and “builds two containers” may be misleading (compose might pull images). Suggest “starts two containers” and keep product names consistent.
-When using llama-stack as a separate service, the existing `docker-compose.yaml` provides the complete setup. This builds two containers for lightspeed core and llama stack. +When using Llama Stack as a separate service, the existing `docker-compose.yaml` provides the complete setup. This starts two containers: lightspeed-core and Llama Stack.
541-541: Naming consistency: use “Llama Stack” (no hyphen) in this heading tooAlign with earlier sections that use “Llama Stack”.
-### Llama-Stack as Library (Library Mode) +### Llama Stack as Library (Library Mode)
543-544: Remove duplicated “directly” and align namingMinor readability fix and naming consistency.
-When embedding llama-stack directly in the container, use the existing `Containerfile` directly (this will not build the llama stack service in a separate container). First modify the `lightspeed-stack.yaml` config to use llama stack in library mode. +When embedding Llama Stack in the container, use the existing `Containerfile` (this will not build the Llama Stack service in a separate container). First modify the `lightspeed-stack.yaml` config to use Llama Stack in library mode.
566-574: Fix platform name capitalization: “macOS”Use “macOS” instead of “macosx”.
-For macosx users: +For macOS users:
576-582: Optional: add a short readiness loop before querying modelsAvoids a race where the service isn’t ready when curl runs.
-```bash -curl -H "Accept: application/json" http://localhost:8080/v1/models -``` +```bash +# Wait for the API to be ready, then list available models +until curl -sf http://localhost:8080/v1/readiness >/dev/null; do + echo "Waiting for lightspeed-core to become ready…" + sleep 1 +done +curl -H "Accept: application/json" http://localhost:8080/v1/models +```
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
💡 Knowledge Base configuration:
- MCP integration is disabled by default for public repositories
- Jira integration is disabled by default for public repositories
- Linear integration is disabled by default for public repositories
You can enable these sources in your CodeRabbit configuration.
📒 Files selected for processing (5)
Containerfile(1 hunks)README.md(1 hunks)lightspeed-stack.yaml(2 hunks)run.yaml(1 hunks)test.containerfile(2 hunks)
🚧 Files skipped from review as they are similar to previous changes (4)
- Containerfile
- run.yaml
- lightspeed-stack.yaml
- test.containerfile
🧰 Additional context used
🧠 Learnings (1)
📚 Learning: 2025-08-18T10:56:55.302Z
Learnt from: matysek
PR: lightspeed-core/lightspeed-stack#292
File: pyproject.toml:0-0
Timestamp: 2025-08-18T10:56:55.302Z
Learning: The lightspeed-stack project intentionally uses a "generic image" approach, bundling many dependencies directly in the base runtime image to work for everyone, rather than using lean base images with optional dependency groups.
Applied to files:
README.md
🪛 LanguageTool
README.md
[grammar] ~516-~516: There might be a mistake here.
Context: ...within lightspeed-core ### Llama-Stack as Separate Service (Server Mode) When us...
(QB_NEW_EN)
[grammar] ~566-~566: Ensure spelling is correct
Context: ... \ my-lightspeed-core:latest For macosx users:bash podman run \ -p 8080:8...
(QB_NEW_EN_ORTHOGRAPHY_ERROR_IDS_1)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (2)
- GitHub Check: build-pr
- GitHub Check: e2e_tests
🔇 Additional comments (3)
README.md (3)
509-515: Good addition: clear two-mode overview for container buildsThe new “Building Container Images” section cleanly separates Server vs Library modes and fits well with the rest of the README.
522-526: No changes needed:api_keyis a valid LlamaStackConfiguration fieldThe
api_keyproperty is defined onLlamaStackConfiguration(src/models/config.py:158–160) and is exercised in both integration (tests/integration/test_configuration.py:60–61) and unit tests (e.g., tests/unit/test_configuration.py:86). The example in README.md correctly reflects the schema.
529-539: Verified compose filename and ports are correctThe repository contains
docker-compose.yamlat the root, and it exposes:
- llama-stack on port 8321
- lightspeed-stack on port 8080
No updates to the README are needed.
umago
left a comment
umago
left a comment
There was a problem hiding this comment.
Thanks, it LGTM! Just a few nitpicks. But since this is docs, I think we need to have a consistent naming for things.
|
|
||
| # no-root user is checked in Konflux | ||
| USER 1001 | ||
| USER 1001 |
| ### Llama-Stack as Separate Service (Server Mode) | ||
| When using llama-stack as a separate service, the existing `docker-compose.yaml` provides the complete setup. This builds two containers for lightspeed core and llama stack. |
There was a problem hiding this comment.
nit:
llama-stack (with hyphen). To keep things consistent, I'm seeing Llama-stack, llama-stack and now llama stack. Let's pick one
|
|
||
| ### Llama-Stack as Library (Library Mode) | ||
|
|
||
| When embedding llama-stack directly in the container, use the existing `Containerfile` directly (this will not build the llama stack service in a separate container). First modify the `lightspeed-stack.yaml` config to use llama stack in library mode. |
There was a problem hiding this comment.
ditto, llama-stack (with hyphen)
4 participants
Description
Type of change
Related Tickets & Documents
Checklist before requesting a review
Testing
Summary by CodeRabbit