Skip to content

fix occasional ttft and tpot metrics test failures #233

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Oct 28, 2025
Merged

fix occasional ttft and tpot metrics test failures #233

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions pkg/llm-d-inference-sim/metrics_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -393,14 +393,14 @@ var _ = Describe("Simulator metrics", Ordered, func() {
modelName := "my_model"
// Send one request, check that ttft and tpot are as defined in the simulator command line params
ctx := context.TODO()
args := []string{"cmd", "--model", modelName, "--mode", common.ModeRandom,
// use mode echo to be sure that response is more than one token - this makes sure that tpot is reported to prometheus
args := []string{"cmd", "--model", modelName, "--mode", common.ModeEcho,
"--time-to-first-token", "200", "--inter-token-latency", "100"}

client, err := startServerWithArgs(ctx, common.ModeRandom, args, nil)
Expect(err).NotTo(HaveOccurred())

openaiclient, params := getOpenAIClientAndChatParams(client, modelName, userMessage, false)
params.MaxTokens = openai.Int(5)

var reqWg, metricsWg sync.WaitGroup
metricsWg.Add(1)
Expand Down Expand Up @@ -451,7 +451,7 @@ var _ = Describe("Simulator metrics", Ordered, func() {
Expect(metrics).To(ContainSubstring("vllm:time_to_first_token_seconds_bucket{model_name=\"my_model\",le=\"640\"} 1"))
Expect(metrics).To(ContainSubstring("vllm:time_to_first_token_seconds_bucket{model_name=\"my_model\",le=\"2560\"} 1"))
Expect(metrics).To(ContainSubstring("vllm:time_to_first_token_seconds_bucket{model_name=\"my_model\",le=\"+Inf\"} 1"))
// tpot
// check tpot only is it exists in metrics, when a single
Expect(metrics).To(ContainSubstring("vllm:time_per_output_token_seconds_bucket{model_name=\"my_model\",le=\"0.01\"} 0"))
Expect(metrics).To(ContainSubstring("vllm:time_per_output_token_seconds_bucket{model_name=\"my_model\",le=\"0.025\"} 0"))
Expect(metrics).To(ContainSubstring("vllm:time_per_output_token_seconds_bucket{model_name=\"my_model\",le=\"0.05\"} 0"))
Expand Down