Skip to content

Commit

Permalink
add containers check in LLM UI test suite
Browse files Browse the repository at this point in the history
  • Loading branch information
@bdattoma
bdattoma committed Apr 24, 2024
1 parent f529dc4 commit 6296e26
Showing 1 changed file with 36 additions and 28 deletions.
64 changes: 36 additions & 28 deletions ...ci/tests/Tests/400__ods_dashboard/420__model_serving/LLMs/422__model_serving_llm_UI.robot
View file
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,7 @@ ${FLAN_MODEL_S3_DIR}= flan-t5-small/flan-t5-small-caikit
${FLAN_GRAMMAR_MODEL_S3_DIR}= flan-t5-large-grammar-synthesis-caikit/flan-t5-large-grammar-synthesis-caikit
${FLAN_LARGE_MODEL_S3_DIR}= flan-t5-large/flan-t5-large
${BLOOM_MODEL_S3_DIR}= bloom-560m/bloom-560m-caikit
${CAIKIT_TGIS_RUNTIME_NAME}= caikit-tgis-runtime


*** Test Cases ***
Expand All @@ -35,8 +36,8 @@ Verify User Can Serve And Query A Model Using The UI
${test_namespace}= Set Variable ${TEST_NS}
${flan_model_name}= Set Variable flan-t5-small-caikit
Deploy Kserve Model Via UI ${flan_model_name} Caikit kserve-connection flan-t5-small/${flan_model_name}
Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace}
Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace} runtime=${CAIKIT_TGIS_RUNTIME_NAME}
Query Model Multiple Times model_name=${flan_model_name}
... inference_type=all-tokens n_times=1
... namespace=${test_namespace} protocol=http
Expand All @@ -55,12 +56,12 @@ Verify User Can Deploy Multiple Models In The Same Namespace Using The UI # rob
${model_two_name}= Set Variable flan-t5-small-caikit
Deploy Kserve Model Via UI ${model_one_name} Caikit kserve-connection
... ${BLOOM_MODEL_S3_DIR}
Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${model_one_name}
... namespace=${test_namespace}
Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${model_one_name}
... namespace=${test_namespace} runtime=${CAIKIT_TGIS_RUNTIME_NAME}
Deploy Kserve Model Via UI ${model_two_name} Caikit kserve-connection
... flan-t5-small/${model_two_name}
Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${model_two_name}
... namespace=${test_namespace}
Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${model_two_name}
... namespace=${test_namespace} runtime=${CAIKIT_TGIS_RUNTIME_NAME}
Query Model Multiple Times inference_type=all-tokens model_name=${model_one_name}
... n_times=5 namespace=${test_namespace} protocol=http
Query Model Multiple Times inference_type=all-tokens model_name=${model_two_name}
Expand All @@ -79,16 +80,16 @@ Verify User Can Deploy Multiple Models In Different Namespaces Using The UI # r
${model_two_name}= Set Variable flan-t5-small-caikit
Deploy Kserve Model Via UI ${model_one_name} Caikit kserve-connection
... bloom-560m/${model_one_name}
Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${model_one_name}
... namespace=singlemodel-multi1
Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${model_one_name}
... namespace=singlemodel-multi1 runtime=${CAIKIT_TGIS_RUNTIME_NAME}
Query Model Multiple Times inference_type=all-tokens model_name=${model_one_name}
... n_times=2 namespace=singlemodel-multi1 protocol=http
Open Data Science Projects Home Page
Set Up Project namespace=singlemodel-multi2 single_prj=${FALSE} dc_name=kserve-connection-2
Deploy Kserve Model Via UI ${model_two_name} Caikit kserve-connection-2
... flan-t5-small/${model_two_name}
Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${model_two_name}
... namespace=singlemodel-multi2
Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${model_two_name}
... namespace=singlemodel-multi2 runtime=${CAIKIT_TGIS_RUNTIME_NAME}
Query Model Multiple Times inference_type=all-tokens model_name=${model_two_name}
... n_times=2 namespace=singlemodel-multi2 protocol=http
[Teardown] Clean Up DSP Page
Expand All @@ -115,8 +116,8 @@ Verify User Can Set Requests And Limits For A Model Using The UI # robocop: dis
${requests}= Create Dictionary cpu=1 memory=4Gi
${limits}= Create Dictionary cpu=2 memory=8Gi
Deploy Kserve Model Via UI ${flan_model_name} Caikit kserve-connection flan-t5-small/${flan_model_name}
Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace}
Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace} runtime=${CAIKIT_TGIS_RUNTIME_NAME}
# ${rev_id}= Get Current Revision ID model_name=${flan_model_name}
# ... namespace=${test_namespace}
Query Model Multiple Times inference_type=all-tokens model_name=${flan_model_name}
Expand All @@ -133,8 +134,8 @@ Verify User Can Set Requests And Limits For A Model Using The UI # robocop: dis
#### Editing the size of an existing model does not work in 2.5, deploying a different one with different size
Deploy Kserve Model Via UI ${flan_model_name}-medium Caikit kserve-connection
... flan-t5-small/${flan_model_name} size=Medium
# Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
# ... namespace=${test_namespace} exp_replicas=1
# Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
# ... namespace=${test_namespace} exp_replicas=1 runtime=${CAIKIT_TGIS_RUNTIME_NAME}
##### Usually our clusters won't have enough resource to actually spawn this, don't wait for pods to be ready
Sleep 5
Container Hardware Resources Should Match Expected container_name=kserve-container
Expand All @@ -153,8 +154,8 @@ Verify Model Can Be Served And Query On A GPU Node Using The UI # robocop: disa
${limits}= Create Dictionary nvidia.com/gpu=1
Deploy Kserve Model Via UI ${model_name} Caikit kserve-connection
... flan-t5-small/${model_name} no_gpus=${1}
Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${model_name}
... namespace=${test_namespace}
Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${model_name}
... namespace=${test_namespace} runtime=${CAIKIT_TGIS_RUNTIME_NAME}
Container Hardware Resources Should Match Expected container_name=kserve-container
... pod_label_selector=serving.kserve.io/inferenceservice=${model_name}
... namespace=${test_namespace} exp_requests=${requests} exp_limits=${limits}
Expand All @@ -176,8 +177,8 @@ Verify Non Admin Can Serve And Query A Model Using The UI # robocop: disable
${test_namespace}= Set Variable non-admin-test
${flan_model_name}= Set Variable flan-t5-small-caikit
Deploy Kserve Model Via UI ${flan_model_name} Caikit kserve-connection flan-t5-small/${flan_model_name}
Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace}
Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace} runtime=${CAIKIT_TGIS_RUNTIME_NAME}
# ${host}= Get KServe Inference Host Via CLI isvc_name=${flan_model_name} namespace=${test_namespace}
# ${body}= Set Variable '{"text": "${EXP_RESPONSES}[queries][0][query_text]"}'
# ${header}= Set Variable 'mm-model-id: ${flan_model_name}'
Expand All @@ -197,8 +198,8 @@ Verify User Can Serve And Query Flan-t5 Grammar Syntax Corrector Using The UI #
${flan_model_name}= Set Variable flan-t5-large-grammar-synthesis-caikit
Deploy Kserve Model Via UI ${flan_model_name} Caikit
... kserve-connection flan-t5-large-grammar-synthesis-caikit/${flan_model_name}
Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace}
Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace} runtime=${CAIKIT_TGIS_RUNTIME_NAME}
Sleep 30s
Query Model Multiple Times inference_type=all-tokens model_name=${flan_model_name}
... n_times=1 namespace=${test_namespace} query_idx=${1} protocol=http
Expand All @@ -215,8 +216,8 @@ Verify User Can Serve And Query Flan-t5 Large Using The UI # robocop: disable
${flan_model_name}= Set Variable flan-t5-large
Deploy Kserve Model Via UI ${flan_model_name} Caikit
... kserve-connection flan-t5-large/flan-t5-large
Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace}
Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace} runtime=${CAIKIT_TGIS_RUNTIME_NAME}
Sleep 30s
Query Model Multiple Times model_name=${flan_model_name}
... inference_type=all-tokens n_times=1
Expand All @@ -239,8 +240,8 @@ Verify User Can Access Model Metrics From UWM Using The UI # robocop: disable
${thanos_url}= Get OpenShift Thanos URL
${token}= Generate Thanos Token
Deploy Kserve Model Via UI ${flan_model_name} Caikit kserve-connection flan-t5-small/${flan_model_name}
Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace}
Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace} runtime=${CAIKIT_TGIS_RUNTIME_NAME}
Wait Until Keyword Succeeds 30 times 4s
... TGI Caikit And Istio Metrics Should Exist thanos_url=${thanos_url} thanos_token=${token}
Query Model Multiple Times model_name=${flan_model_name}
Expand Down Expand Up @@ -284,8 +285,8 @@ Verify User With Edit Permission Can Deploy Query And Delete A LLM
IF ${modal}==${TRUE}
Click Element //button[@aria-label="Close"]
END
Run Keyword And Continue On Failure Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace}
Run Keyword And Continue On Failure Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace} runtime=${CAIKIT_TGIS_RUNTIME_NAME}
Run Keyword And Continue On Failure Query Model Multiple Times model_name=${flan_model_name}
... inference_type=all-tokens n_times=1
... namespace=${test_namespace} protocol=http
Expand All @@ -312,8 +313,8 @@ Verify User With Admin Permission Can Deploy Query And Delete A LLM
Login To RHODS Dashboard ${TEST_USER_3.USERNAME} ${TEST_USER_3.PASSWORD} ${TEST_USER_3.AUTH_TYPE}
Wait For RHODS Dashboard To Load expected_page=${test_namespace} wait_for_cards=${FALSE}
Deploy Kserve Model Via UI ${flan_model_name} Caikit kserve-connection flan-t5-small/${flan_model_name}
Wait For Pods To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace}
Wait For Model KServe Deployment To Be Ready label_selector=serving.kserve.io/inferenceservice=${flan_model_name}
... namespace=${test_namespace} runtime=${CAIKIT_TGIS_RUNTIME_NAME}
Query Model Multiple Times model_name=${flan_model_name}
... inference_type=all-tokens n_times=1
... namespace=${test_namespace} protocol=http
Expand All @@ -339,3 +340,10 @@ Setup Kserve UI Test
Load Expected Responses
Launch Dashboard ${user} ${pw} ${auth} ${ODH_DASHBOARD_URL} ${BROWSER.NAME} ${BROWSER.OPTIONS}
Fetch CA Certificate If RHODS Is Self-Managed
${dsc_kserve_mode}= Get KServe Default Deployment Mode From DSC
Set Suite Variable ${DSC_KSERVE_MODE} ${dsc_kserve_mode}
IF "${dsc_kserve_mode}" == "RawDeployment"
Set Suite Variable ${IS_KSERVE_RAW} ${TRUE}
ELSE
Set Suite Variable ${IS_KSERVE_RAW} ${FALSE}
END

0 comments on commit 6296e26

Please sign in to comment.