Skip to content

Commit 658fb2c

Browse files
ashwinbashwinb
committed
refactor(k8s): update run configs to v2 storage and registered_resources structure
Migrates k8s run configs to match the updated run configs - Replace storage.references with storage.stores - Wrap resources under registered_resources section - Update provider configs to use persistence with namespace/backend - Add telemetry and vector_stores top-level sections - Simplify agent/files metadata store configuration
1 parent 0e57233 commit 658fb2c

File tree

2 files changed

+107
-99
lines changed

2 files changed

+107
-99
lines changed

‎benchmarking/k8s-benchmark/stack_run_config.yaml‎

Lines changed: 52 additions & 48 deletions
Original file line numberDiff line numberDiff line change
@@ -27,28 +27,24 @@ providers:
2727
config:
2828
storage_dir: ${env.FILES_STORAGE_DIR:=~/.llama/distributions/starter/files}
2929
metadata_store:
30-
type: sqlite
31-
db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/starter}/files_metadata.db
30+
table_name: files_metadata
31+
backend: sql_default
3232
vector_io:
3333
- provider_id: ${env.ENABLE_CHROMADB:+chromadb}
3434
provider_type: remote::chromadb
3535
config:
3636
url: ${env.CHROMADB_URL:=}
37-
kvstore:
38-
type: postgres
39-
host: ${env.POSTGRES_HOST:=localhost}
40-
port: ${env.POSTGRES_PORT:=5432}
41-
db: ${env.POSTGRES_DB:=llamastack}
42-
user: ${env.POSTGRES_USER:=llamastack}
43-
password: ${env.POSTGRES_PASSWORD:=llamastack}
37+
persistence:
38+
namespace: vector_io::chroma_remote
39+
backend: kv_default
4440
files:
4541
- provider_id: meta-reference-files
4642
provider_type: inline::localfs
4743
config:
4844
storage_dir: ${env.FILES_STORAGE_DIR:=~/.llama/distributions/starter/files}
4945
metadata_store:
50-
type: sqlite
51-
db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/starter}/files_metadata.db
46+
table_name: files_metadata
47+
backend: sql_default
5248
safety:
5349
- provider_id: llama-guard
5450
provider_type: inline::llama-guard
@@ -58,20 +54,15 @@ providers:
5854
- provider_id: meta-reference
5955
provider_type: inline::meta-reference
6056
config:
61-
persistence_store:
62-
type: postgres
63-
host: ${env.POSTGRES_HOST:=localhost}
64-
port: ${env.POSTGRES_PORT:=5432}
65-
db: ${env.POSTGRES_DB:=llamastack}
66-
user: ${env.POSTGRES_USER:=llamastack}
67-
password: ${env.POSTGRES_PASSWORD:=llamastack}
68-
responses_store:
69-
type: postgres
70-
host: ${env.POSTGRES_HOST:=localhost}
71-
port: ${env.POSTGRES_PORT:=5432}
72-
db: ${env.POSTGRES_DB:=llamastack}
73-
user: ${env.POSTGRES_USER:=llamastack}
74-
password: ${env.POSTGRES_PASSWORD:=llamastack}
57+
persistence:
58+
agent_state:
59+
namespace: agents
60+
backend: kv_default
61+
responses:
62+
table_name: responses
63+
backend: sql_default
64+
max_write_queue_size: 10000
65+
num_writers: 4
7566
telemetry:
7667
- provider_id: meta-reference
7768
provider_type: inline::meta-reference
@@ -112,32 +103,45 @@ storage:
112103
db: ${env.POSTGRES_DB:=llamastack}
113104
user: ${env.POSTGRES_USER:=llamastack}
114105
password: ${env.POSTGRES_PASSWORD:=llamastack}
115-
references:
106+
stores:
116107
metadata:
117-
backend: kv_default
118108
namespace: registry
109+
backend: kv_default
119110
inference:
120-
backend: sql_default
121111
table_name: inference_store
122-
models:
123-
- metadata:
124-
embedding_dimension: 768
125-
model_id: nomic-embed-text-v1.5
126-
provider_id: sentence-transformers
127-
model_type: embedding
128-
- model_id: ${env.INFERENCE_MODEL}
129-
provider_id: vllm-inference
130-
model_type: llm
131-
shields:
132-
- shield_id: ${env.SAFETY_MODEL:=meta-llama/Llama-Guard-3-1B}
133-
vector_dbs: []
134-
datasets: []
135-
scoring_fns: []
136-
benchmarks: []
137-
tool_groups:
138-
- toolgroup_id: builtin::websearch
139-
provider_id: tavily-search
140-
- toolgroup_id: builtin::rag
141-
provider_id: rag-runtime
112+
backend: sql_default
113+
max_write_queue_size: 10000
114+
num_writers: 4
115+
conversations:
116+
table_name: openai_conversations
117+
backend: sql_default
118+
registered_resources:
119+
models:
120+
- metadata:
121+
embedding_dimension: 768
122+
model_id: nomic-embed-text-v1.5
123+
provider_id: sentence-transformers
124+
model_type: embedding
125+
- model_id: ${env.INFERENCE_MODEL}
126+
provider_id: vllm-inference
127+
model_type: llm
128+
shields:
129+
- shield_id: ${env.SAFETY_MODEL:=meta-llama/Llama-Guard-3-1B}
130+
vector_dbs: []
131+
datasets: []
132+
scoring_fns: []
133+
benchmarks: []
134+
tool_groups:
135+
- toolgroup_id: builtin::websearch
136+
provider_id: tavily-search
137+
- toolgroup_id: builtin::rag
138+
provider_id: rag-runtime
142139
server:
143140
port: 8323
141+
telemetry:
142+
enabled: true
143+
vector_stores:
144+
default_provider_id: chromadb
145+
default_embedding_model:
146+
provider_id: sentence-transformers
147+
model_id: nomic-ai/nomic-embed-text-v1.5

‎docs/docs/distributions/k8s/stack_run_config.yaml‎

Lines changed: 55 additions & 51 deletions
Original file line numberDiff line numberDiff line change
@@ -32,21 +32,17 @@ providers:
3232
provider_type: remote::chromadb
3333
config:
3434
url: ${env.CHROMADB_URL:=}
35-
kvstore:
36-
type: postgres
37-
host: ${env.POSTGRES_HOST:=localhost}
38-
port: ${env.POSTGRES_PORT:=5432}
39-
db: ${env.POSTGRES_DB:=llamastack}
40-
user: ${env.POSTGRES_USER:=llamastack}
41-
password: ${env.POSTGRES_PASSWORD:=llamastack}
35+
persistence:
36+
namespace: vector_io::chroma_remote
37+
backend: kv_default
4238
files:
4339
- provider_id: meta-reference-files
4440
provider_type: inline::localfs
4541
config:
4642
storage_dir: ${env.FILES_STORAGE_DIR:=~/.llama/distributions/starter/files}
4743
metadata_store:
48-
type: sqlite
49-
db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/starter}/files_metadata.db
44+
table_name: files_metadata
45+
backend: sql_default
5046
safety:
5147
- provider_id: llama-guard
5248
provider_type: inline::llama-guard
@@ -56,20 +52,15 @@ providers:
5652
- provider_id: meta-reference
5753
provider_type: inline::meta-reference
5854
config:
59-
persistence_store:
60-
type: postgres
61-
host: ${env.POSTGRES_HOST:=localhost}
62-
port: ${env.POSTGRES_PORT:=5432}
63-
db: ${env.POSTGRES_DB:=llamastack}
64-
user: ${env.POSTGRES_USER:=llamastack}
65-
password: ${env.POSTGRES_PASSWORD:=llamastack}
66-
responses_store:
67-
type: postgres
68-
host: ${env.POSTGRES_HOST:=localhost}
69-
port: ${env.POSTGRES_PORT:=5432}
70-
db: ${env.POSTGRES_DB:=llamastack}
71-
user: ${env.POSTGRES_USER:=llamastack}
72-
password: ${env.POSTGRES_PASSWORD:=llamastack}
55+
persistence:
56+
agent_state:
57+
namespace: agents
58+
backend: kv_default
59+
responses:
60+
table_name: responses
61+
backend: sql_default
62+
max_write_queue_size: 10000
63+
num_writers: 4
7364
telemetry:
7465
- provider_id: meta-reference
7566
provider_type: inline::meta-reference
@@ -110,40 +101,53 @@ storage:
110101
db: ${env.POSTGRES_DB:=llamastack}
111102
user: ${env.POSTGRES_USER:=llamastack}
112103
password: ${env.POSTGRES_PASSWORD:=llamastack}
113-
references:
104+
stores:
114105
metadata:
115-
backend: kv_default
116106
namespace: registry
107+
backend: kv_default
117108
inference:
118-
backend: sql_default
119109
table_name: inference_store
120-
models:
121-
- metadata:
122-
embedding_dimension: 768
123-
model_id: nomic-embed-text-v1.5
124-
provider_id: sentence-transformers
125-
model_type: embedding
126-
- metadata: {}
127-
model_id: ${env.INFERENCE_MODEL}
128-
provider_id: vllm-inference
129-
model_type: llm
130-
- metadata: {}
131-
model_id: ${env.SAFETY_MODEL:=meta-llama/Llama-Guard-3-1B}
132-
provider_id: vllm-safety
133-
model_type: llm
134-
shields:
135-
- shield_id: ${env.SAFETY_MODEL:=meta-llama/Llama-Guard-3-1B}
136-
vector_dbs: []
137-
datasets: []
138-
scoring_fns: []
139-
benchmarks: []
140-
tool_groups:
141-
- toolgroup_id: builtin::websearch
142-
provider_id: tavily-search
143-
- toolgroup_id: builtin::rag
144-
provider_id: rag-runtime
110+
backend: sql_default
111+
max_write_queue_size: 10000
112+
num_writers: 4
113+
conversations:
114+
table_name: openai_conversations
115+
backend: sql_default
116+
registered_resources:
117+
models:
118+
- metadata:
119+
embedding_dimension: 768
120+
model_id: nomic-embed-text-v1.5
121+
provider_id: sentence-transformers
122+
model_type: embedding
123+
- metadata: {}
124+
model_id: ${env.INFERENCE_MODEL}
125+
provider_id: vllm-inference
126+
model_type: llm
127+
- metadata: {}
128+
model_id: ${env.SAFETY_MODEL:=meta-llama/Llama-Guard-3-1B}
129+
provider_id: vllm-safety
130+
model_type: llm
131+
shields:
132+
- shield_id: ${env.SAFETY_MODEL:=meta-llama/Llama-Guard-3-1B}
133+
vector_dbs: []
134+
datasets: []
135+
scoring_fns: []
136+
benchmarks: []
137+
tool_groups:
138+
- toolgroup_id: builtin::websearch
139+
provider_id: tavily-search
140+
- toolgroup_id: builtin::rag
141+
provider_id: rag-runtime
145142
server:
146143
port: 8321
147144
auth:
148145
provider_config:
149146
type: github_token
147+
telemetry:
148+
enabled: true
149+
vector_stores:
150+
default_provider_id: chromadb
151+
default_embedding_model:
152+
provider_id: sentence-transformers
153+
model_id: nomic-ai/nomic-embed-text-v1.5

0 commit comments

Comments
 (0)