You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This section describes how to use FluxCD for GitOps-based deployment of Dynamo inference graphs. GitOps enables you to manage your Dynamo deployments declaratively using Git as the source of truth. We'll use the [aggregated vLLM example](../../../components/backends/vllm/README.md) to demonstrate the workflow.
38
-
39
-
### Prerequisites
40
-
41
-
- A Kubernetes cluster with [Dynamo Cloud](installation_guide.md) installed
42
-
-[FluxCD](https://fluxcd.io/flux/installation/) installed in your cluster
43
-
- A Git repository to store your deployment configurations
44
-
45
-
### Workflow Overview
46
-
47
-
The GitOps workflow for Dynamo deployments consists of three main steps:
48
-
49
-
1. Build and push the Dynamo Operator
50
-
2. Create and commit a DynamoGraphDeployment custom resource for initial deployment
51
-
3. Update the graph by building a new version and updating the CR for subsequent updates
52
-
53
-
### Step 1: Build and Push Dynamo Cloud Operator
54
-
55
-
First, follow to [See Install Dynamo Cloud](README.md).
56
-
57
-
### Step 2: Create Initial Deployment
58
-
59
-
Create a new file in your Git repository (e.g., `deployments/llm-agg.yaml`) with the following content:
60
-
61
-
```yaml
62
-
apiVersion: nvidia.com/v1alpha1
63
-
kind: DynamoGraphDeployment
64
-
metadata:
65
-
name: llm-agg
66
-
spec:
67
-
services:
68
-
Frontend:
69
-
replicas: 1
70
-
envs:
71
-
- name: SPECIFIC_ENV_VAR
72
-
value: some_specific_value
73
-
Processor:
74
-
replicas: 1
75
-
envs:
76
-
- name: SPECIFIC_ENV_VAR
77
-
value: some_specific_value
78
-
VllmWorker:
79
-
replicas: 1
80
-
envs:
81
-
- name: SPECIFIC_ENV_VAR
82
-
value: some_specific_value
83
-
# Add PVC for model storage
84
-
pvc:
85
-
name: vllm-model-storage
86
-
mountPath: /models
87
-
size: 100Gi
88
-
```
89
-
90
-
Commit and push this file to your Git repository. FluxCD will detect the new CR and create the initial deployment in your cluster. The operator will:
91
-
- Create the specified PVCs
92
-
- Build container images for all components
93
-
- Deploy the services with the configured resources
94
-
95
-
### Step 3: Update Existing Deployment
96
-
97
-
To update your pipeline, just update the associated DynamoGraphDeployment CRD
98
-
99
-
The Dynamo operator will automatically reconcile it.
This section describes how to use FluxCD for GitOps-based deployment of Dynamo inference graphs. GitOps enables you to manage your Dynamo deployments declaratively using Git as the source of truth. We'll use the [aggregated vLLM example](../../../components/backends/vllm/README.md) to demonstrate the workflow.
4
+
5
+
## Prerequisites
6
+
7
+
- A Kubernetes cluster with [Dynamo Cloud](/docs/guides/dynamo_deploy/installation_guide.md) installed
8
+
-[FluxCD](https://fluxcd.io/flux/installation/) installed in your cluster
9
+
- A Git repository to store your deployment configurations
10
+
11
+
## Workflow Overview
12
+
13
+
The GitOps workflow for Dynamo deployments consists of three main steps:
14
+
15
+
1. Build and push the Dynamo Operator
16
+
2. Create and commit a DynamoGraphDeployment custom resource for initial deployment
17
+
3. Update the graph by building a new version and updating the CR for subsequent updates
18
+
19
+
## Step 1: Build and Push Dynamo Cloud Operator
20
+
21
+
First, follow to [See Install Dynamo Cloud](/docs/guides/dynamo_deploy/installation_guide.md).
22
+
23
+
## Step 2: Create Initial Deployment
24
+
25
+
Create a new file in your Git repository (e.g., `deployments/llm-agg.yaml`) with the following content:
26
+
27
+
```yaml
28
+
apiVersion: nvidia.com/v1alpha1
29
+
kind: DynamoGraphDeployment
30
+
metadata:
31
+
name: llm-agg
32
+
spec:
33
+
services:
34
+
Frontend:
35
+
replicas: 1
36
+
envs:
37
+
- name: SPECIFIC_ENV_VAR
38
+
value: some_specific_value
39
+
Processor:
40
+
replicas: 1
41
+
envs:
42
+
- name: SPECIFIC_ENV_VAR
43
+
value: some_specific_value
44
+
VllmWorker:
45
+
replicas: 1
46
+
envs:
47
+
- name: SPECIFIC_ENV_VAR
48
+
value: some_specific_value
49
+
# Add PVC for model storage
50
+
pvc:
51
+
name: vllm-model-storage
52
+
mountPath: /models
53
+
size: 100Gi
54
+
```
55
+
56
+
Commit and push this file to your Git repository. FluxCD will detect the new CR and create the initial Dynamo deployment in your cluster.
57
+
58
+
## Step 3: Update Existing Deployment
59
+
60
+
To update your pipeline, just update the associated DynamoGraphDeployment CRD
61
+
62
+
The Dynamo operator will automatically reconcile it.
0 commit comments