Skip to content

Same agent Smaller bill. The tools and infrastructure to get more work from your AI. Fewer tokens. Less compute. Lower cost.

Add `platform` and `generation` labels when exporting to New Relic and DataDog (#58)

Patcher99e0dc16ecc302
diff --git a/README.md b/README.md
index c33a2904..86d8d7cf 100644
--- a/README.md
+++ b/README.md
@@ -25,7 +25,7 @@ Leveraging Doku, you can get:
- πŸ•΅οΈ **In-depth LLM Monitoring**: Track every request to LLM platforms like OpenAI with precision, ensuring comprehensive visibility over your model's interactions.
- πŸŽ›οΈ **Granular Usage Insights of your LLM Applications**: Assess your LLM's performance and costs with fine-grained control, breaking down metrics by environment (such as staging or production) or application, to optimize for efficiency and scalability.
-- πŸ“ˆ **Connect to Observability Platforms**: Export LLM Observablity data and insights from Doku to popular observability platforms such as Grafana Cloud or Datadog.
+- πŸ“ˆ **Connect to Observability Platforms**: Export LLM Observablity data and insights from Doku to popular observability platforms such as Grafana OSS, Grafana Cloud, Datadog and New Relic.
- πŸ‘₯ **Team-centric Workflow**: Enhance team collaboration with seamless data sharing capabilities, creating an integrated environment for observability-driven teamwork.
## Supported LLMs:
diff --git a/src/ingester/connections/connections.go b/src/ingester/connections/connections.go
index 57d2e540..d01fc836 100644
--- a/src/ingester/connections/connections.go
+++ b/src/ingester/connections/connections.go
@@ -6,8 +6,6 @@ import (
"regexp"
"strings"
"time"
-
- "github.com/rs/zerolog/log"
)
var (
@@ -91,7 +89,5 @@ func SendToPlatform(data map[string]interface{}) {
configureNewRelicData(data)
} else if dataDogMetricsUrl != "" {
configureDataDogData(data)
- } else {
- log.Info().Msg("No Observability Platform configured")
}
}
diff --git a/src/ingester/connections/datadog.go b/src/ingester/connections/datadog.go
index 1bcbd6c9..cdfaacfc 100644
--- a/src/ingester/connections/datadog.go
+++ b/src/ingester/connections/datadog.go
@@ -13,6 +13,20 @@ import (
func configureDataDogData(data map[string]interface{}) {
currentTime := time.Now().Unix()
+ // Extract the platform from the endpoint string
+ endpointParts := strings.Split(data["endpoint"].(string), ".")
+ if len(endpointParts) > 0 {
+ platform = endpointParts[0] // The first part of the endpoint string is the platform.
+ } else {
+ platform = "Unknown" // If the endpoint string is not in the expected format, set the platform to "unknown".
+ }
+
+ // Determine the type based on the endpoint by consulting the mapping.
+ call_type, found := endpointTypeMapping[data["endpoint"].(string)]
+ if !found {
+ call_type = "Unknown"
+ }
+
if data["endpoint"] == "openai.chat.completions" || data["endpoint"] == "openai.completions" || data["endpoint"] == "cohere.generate" || data["endpoint"] == "cohere.chat" || data["endpoint"] == "cohere.summarize" || data["endpoint"] == "anthropic.completions" {
if data["finishReason"] == nil {
data["finishReason"] = "null"
@@ -23,37 +37,37 @@ func configureDataDogData(data map[string]interface{}) {
"metric": "doku.llm.completion.tokens",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "finishReason:%v"]
- }`, currentTime, data["completionTokens"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "finishReason:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["completionTokens"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"], platform, call_type),
fmt.Sprintf(`{
"metric": "doku.llm.prompt.tokens",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "finishReason:%v"]
- }`, currentTime, data["promptTokens"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "finishReason:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["promptTokens"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"], platform, call_type),
fmt.Sprintf(`{
"metric": "doku.llm.total.tokens",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "finishReason:%v"]
- }`, currentTime, data["totalTokens"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "finishReason:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["totalTokens"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"], platform, call_type),
fmt.Sprintf(`{
"metric": "doku.llm.request.duration",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "finishReason:%v"]
- }`, currentTime, data["requestDuration"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "finishReason:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["requestDuration"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"], platform, call_type),
fmt.Sprintf(`{
"metric": "doku.llm.usage.cost",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "finishReason:%v"]
- }`, currentTime, data["usageCost"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "finishReason:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["usageCost"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"], platform, call_type),
}
metrics := fmt.Sprintf(`{"series": [%s]}`, strings.Join(metricStrings, ","))
@@ -66,17 +80,17 @@ func configureDataDogData(data map[string]interface{}) {
fmt.Sprintf(`{
"ddsource": "doku",
"message": "%s",
- "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:prompt",
- "hostname": "doku-ingester",
+ "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:prompt,platform:%v,generation:%v",
+ "hostname": "doku",
"service": "%v"
- }`, normalizeString(data["prompt"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["applicationName"]),
+ }`, normalizeString(data["prompt"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type, data["applicationName"]),
fmt.Sprintf(`{
"ddsource": "doku",
"message": "%s",
- "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:response",
- "hostname": "doku-ingester",
+ "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:response,platform:%v,generation:%v",
+ "hostname": "doku",
"service": "%v"
- }`, normalizeString(data["response"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["applicationName"]),
+ }`, normalizeString(data["response"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type, data["applicationName"]),
}
logs := fmt.Sprintf("[%s]", strings.Join(logEntries, ","))
@@ -92,30 +106,30 @@ func configureDataDogData(data map[string]interface{}) {
"metric": "doku.llm.prompt.tokens",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v"]
- }`, currentTime, data["promptTokens"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["promptTokens"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
fmt.Sprintf(`{
"metric": "doku.llm.total.tokens",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v"]
- }`, currentTime, data["totalTokens"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["totalTokens"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
fmt.Sprintf(`{
"metric": "doku.llm.request.duration",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v"]
- }`, currentTime, data["requestDuration"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["requestDuration"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
fmt.Sprintf(`{
"metric": "doku.llm.usage.cost",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v"]
- }`, currentTime, data["usageCost"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["usageCost"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
}
metrics := fmt.Sprintf(`{"series": [%s]}`, strings.Join(metricStrings, ","))
@@ -128,10 +142,10 @@ func configureDataDogData(data map[string]interface{}) {
fmt.Sprintf(`{
"ddsource": "doku",
"message": "%s",
- "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:prompt",
- "hostname": "doku-ingester",
+ "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:prompt,platform:%v,generation:%v",
+ "hostname": "doku",
"service": "%v"
- }`, normalizeString(data["prompt"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["applicationName"]),
+ }`, normalizeString(data["prompt"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type, data["applicationName"]),
}
logs := fmt.Sprintf("[%s]", strings.Join(logEntries, ","))
@@ -145,23 +159,23 @@ func configureDataDogData(data map[string]interface{}) {
"metric": "doku.llm.prompt.tokens",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v"]
- }`, currentTime, data["promptTokens"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["promptTokens"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
fmt.Sprintf(`{
"metric": "doku.llm.request.duration",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v"]
- }`, currentTime, data["requestDuration"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["requestDuration"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
fmt.Sprintf(`{
"metric": "doku.llm.usage.cost",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v"]
- }`, currentTime, data["usageCost"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["usageCost"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
}
metrics := fmt.Sprintf(`{"series": [%s]}`, strings.Join(metricStrings, ","))
@@ -174,10 +188,10 @@ func configureDataDogData(data map[string]interface{}) {
fmt.Sprintf(`{
"ddsource": "doku",
"message": "%s",
- "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:prompt",
- "hostname": "doku-ingester",
+ "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:prompt,platform:%v,generation:%v",
+ "hostname": "doku",
"service": "%v"
- }`, normalizeString(data["prompt"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["applicationName"]),
+ }`, normalizeString(data["prompt"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type, data["applicationName"]),
}
logs := fmt.Sprintf("[%s]", strings.Join(logEntries, ","))
@@ -192,9 +206,9 @@ func configureDataDogData(data map[string]interface{}) {
"metric": "doku.llm.request.duration",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v"]
- }`, currentTime, data["requestDuration"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["requestDuration"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
}
metrics := fmt.Sprintf(`{"series": [%s]}`, strings.Join(metricStrings, ","))
@@ -208,16 +222,16 @@ func configureDataDogData(data map[string]interface{}) {
"metric": "doku.llm.request.duration",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "imageSize:%v","imageQuality:%v"]
- }`, currentTime, data["requestDuration"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["imageSize"], data["imageQuality"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "imageSize:%v","imageQuality:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["requestDuration"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["imageSize"], data["imageQuality"], platform, call_type),
fmt.Sprintf(`{
"metric": "doku.llm.usage.cost",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "imageSize:%v","imageQuality:%v"]
- }`, currentTime, data["usageCost"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["imageSize"], data["imageQuality"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "imageSize:%v","imageQuality:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["usageCost"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["imageSize"], data["imageQuality"], platform, call_type),
}
metrics := fmt.Sprintf(`{"series": [%s]}`, strings.Join(metricStrings, ","))
@@ -242,20 +256,20 @@ func configureDataDogData(data map[string]interface{}) {
logEntries = append(logEntries, fmt.Sprintf(`{
"ddsource": "doku",
"message": "%s",
- "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:prompt",
- "hostname": "doku-ingester",
+ "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:prompt,platform:%v,generation:%v",
+ "hostname": "doku",
"service": "%v"
- }`, promptMessage, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["applicationName"]),
+ }`, promptMessage, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type, data["applicationName"]),
)
}
logEntries = append(logEntries, fmt.Sprintf(`{
"ddsource": "doku",
"message": "%s",
- "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:image",
- "hostname": "doku-ingester",
+ "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:image,platform:%v,generation:%v",
+ "hostname": "doku",
"service": "%v"
- }`, data["image"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["applicationName"]),
+ }`, data["image"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type, data["applicationName"]),
)
logs := fmt.Sprintf("[%s]", strings.Join(logEntries, ","))
@@ -269,16 +283,16 @@ func configureDataDogData(data map[string]interface{}) {
"metric": "doku.llm.request.duration",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "audioVoice:%v"]
- }`, currentTime, data["requestDuration"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["audioVoice"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "audioVoice:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["requestDuration"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["audioVoice"], platform, call_type),
fmt.Sprintf(`{
"metric": "doku.llm.usage.cost",
"type": 0,
"points": [{ "timestamp": %d, "value": %f }],
- "resources": [{ "name": "doku-ingester", "type": "host" }],
- "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "audioVoice:%v"]
- }`, currentTime, data["usageCost"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["audioVoice"]),
+ "resources": [{ "name": "doku", "type": "host" }],
+ "tags": ["environment:%v", "endpoint:%v", "applicationName:%v", "source:%v", "model:%v", "audioVoice:%v", "platform:%v", "generation:%v"]
+ }`, currentTime, data["usageCost"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["audioVoice"], platform, call_type),
}
metrics := fmt.Sprintf(`{"series": [%s]}`, strings.Join(metricStrings, ","))
@@ -291,10 +305,10 @@ func configureDataDogData(data map[string]interface{}) {
fmt.Sprintf(`{
"ddsource": "doku",
"message": "%s",
- "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:prompt",
- "hostname": "doku-ingester",
+ "ddtags": "environment:%v,endpoint:%v,applicationName:%v,source:%v,model:%v,type:prompt,platform:%v,generation:%v",
+ "hostname": "doku",
"service": "%v"
- }`, normalizeString(data["prompt"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["applicationName"]),
+ }`, normalizeString(data["prompt"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type, data["applicationName"]),
}
logs := fmt.Sprintf("[%s]", strings.Join(logEntries, ","))
diff --git a/src/ingester/connections/grafana-cloud.go b/src/ingester/connections/grafana-cloud.go
index 1b387e9a..a6edc1dc 100644
--- a/src/ingester/connections/grafana-cloud.go
+++ b/src/ingester/connections/grafana-cloud.go
@@ -28,7 +28,6 @@ func configureGrafanaCloudData(data map[string]interface{}) {
if !found {
call_type = "Unknown"
}
- fmt.Println("Platform: ", call_type)
if data["endpoint"] == "openai.chat.completions" || data["endpoint"] == "openai.completions" || data["endpoint"] == "cohere.generate" || data["endpoint"] == "cohere.chat" || data["endpoint"] == "cohere.summarize" || data["endpoint"] == "anthropic.completions" {
if data["finishReason"] == nil {
diff --git a/src/ingester/connections/newrelic.go b/src/ingester/connections/newrelic.go
index 6eeebc37..c6c6e149 100644
--- a/src/ingester/connections/newrelic.go
+++ b/src/ingester/connections/newrelic.go
@@ -15,6 +15,20 @@ func configureNewRelicData(data map[string]interface{}) {
// The current time for the timestamp field.
currentTime := strconv.FormatInt(time.Now().Unix(), 10)
+ // Extract the platform from the endpoint string
+ endpointParts := strings.Split(data["endpoint"].(string), ".")
+ if len(endpointParts) > 0 {
+ platform = endpointParts[0] // The first part of the endpoint string is the platform.
+ } else {
+ platform = "Unknown" // If the endpoint string is not in the expected format, set the platform to "unknown".
+ }
+
+ // Determine the type based on the endpoint by consulting the mapping.
+ call_type, found := endpointTypeMapping[data["endpoint"].(string)]
+ if !found {
+ call_type = "Unknown"
+ }
+
if data["endpoint"] == "openai.chat.completions" || data["endpoint"] == "openai.completions" || data["endpoint"] == "cohere.generate" || data["endpoint"] == "cohere.chat" || data["endpoint"] == "cohere.summarize" || data["endpoint"] == "anthropic.completions" {
if data["finishReason"] == nil {
data["finishReason"] = "null"
@@ -26,36 +40,36 @@ func configureNewRelicData(data map[string]interface{}) {
"type": "gauge",
"value": %f,
"timestamp": %s,
- "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "finishReason": "%v"}
- }`, data["completionTokens"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"]),
+ "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "finishReason": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["completionTokens"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"], platform, call_type),
fmt.Sprintf(`{
"name": "doku.LLM.Prompt.Tokens",
"type": "gauge",
"value": %f,
"timestamp": %s,
- "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "finishReason": "%v"}
- }`, data["promptTokens"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"]),
+ "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "finishReason": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["promptTokens"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"], platform, call_type),
fmt.Sprintf(`{
"name": "doku.LLM.Total.Tokens",
"type": "gauge",
"value": %f,
"timestamp": %s,
- "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "finishReason": "%v"}
- }`, data["totalTokens"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"]),
+ "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "finishReason": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["totalTokens"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"], platform, call_type),
fmt.Sprintf(`{
"name": "doku.LLM.Request.Duration",
"type": "gauge",
"value": %v,
"timestamp": %s,
- "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "finishReason": "%v"}
- }`, data["requestDuration"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"]),
+ "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "finishReason": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["requestDuration"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"], platform, call_type),
fmt.Sprintf(`{
"name": "doku.LLM.Usage.Cost",
"type": "gauge",
"value": %v,
"timestamp": %s,
- "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "finishReason": "%v"}
- }`, data["usageCost"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"]),
+ "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "finishReason": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["usageCost"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["finishReason"], platform, call_type),
}
// Join the individual metric strings into a comma-separated string and enclose in a JSON array.
@@ -77,9 +91,12 @@ func configureNewRelicData(data map[string]interface{}) {
"applicationName": "%v",
"source": "%v",
"model": "%v",
- "type": "response"
+ "type": "response",
+ "platform": "%v",
+ "generation": "%v",
+ "job.name": "doku"
}
- }`, currentTime, normalizeString(data["response"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]), // Assuming 'response' is present in data and is a string.
+ }`, currentTime, normalizeString(data["response"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type), // Assuming 'response' is present in data and is a string.
fmt.Sprintf(`{
"timestamp": %s,
@@ -90,9 +107,12 @@ func configureNewRelicData(data map[string]interface{}) {
"applicationName": "%v",
"source": "%v",
"model": "%v",
- "type": "prompt"
+ "type": "prompt",
+ "platform": "%v",
+ "generation": "%v",
+ "job.name": "doku"
}
- }`, currentTime, normalizeString(data["prompt"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]), // Assuming 'prompt' is present in data and is a string.
+ }`, currentTime, normalizeString(data["prompt"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type), // Assuming 'prompt' is present in data and is a string.
}
// Combine the individual log entries into a full JSON payload
@@ -111,29 +131,29 @@ func configureNewRelicData(data map[string]interface{}) {
"type": "gauge",
"value": %v,
"timestamp": %s,
- "attributes": { "environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v"}
- }`, data["promptTokens"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "attributes": { "environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["promptTokens"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
fmt.Sprintf(`{
"name": "doku.LLM.Total.Tokens",
"type": "gauge",
"value": %v,
"timestamp": %s,
- "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v"}
- }`, data["totalTokens"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["totalTokens"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
fmt.Sprintf(`{
"name": "doku.LLM.Request.Duration",
"type": "gauge",
"value": %v,
"timestamp": %s,
- "attributes": { "environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v"}
- }`, data["requestDuration"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "attributes": { "environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["requestDuration"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
fmt.Sprintf(`{
"name": "doku.LLM.Usage.Cost",
"type": "gauge",
"value": %v,
"timestamp": %s,
- "attributes": { "environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v"}
- }`, data["usageCost"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "attributes": { "environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["usageCost"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
}
// Join the individual metric strings into a comma-separated string and enclose in a JSON array.
@@ -155,9 +175,12 @@ func configureNewRelicData(data map[string]interface{}) {
"applicationName": "%v",
"source": "%v",
"model": "%v",
- "type": "prompt"
+ "type": "prompt",
+ "platform": "%v",
+ "generation": "%v",
+ "job.name": "doku"
}
- }`, currentTime, normalizeString(data["prompt"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ }`, currentTime, normalizeString(data["prompt"].(string)), data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type), // Assuming 'prompt' is present in data and is a string.
}
// Combine the individual log entries into a full JSON payload
jsonData = fmt.Sprintf(`[{"logs": [%s]}]`, strings.Join(logs, ","))
@@ -173,22 +196,22 @@ func configureNewRelicData(data map[string]interface{}) {
"type": "gauge",
"value": %v,
"timestamp": %s,
- "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v"}
- }`, data["promptTokens"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["promptTokens"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
fmt.Sprintf(`{
"name": "doku.LLM.Request.Duration",
"type": "gauge",
"value": %v,
"timestamp": %s,
- "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v"}
- }`, data["requestDuration"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["requestDuration"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
fmt.Sprintf(`{
"name": "doku.LLM.Usage.Cost",
"type": "gauge",
"value": %v,
"timestamp": %s,
- "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v"}
- }`, data["usageCost"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["usageCost"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
}
// Join the individual metric strings into a comma-separated string and enclose in a JSON array.
@@ -210,7 +233,10 @@ func configureNewRelicData(data map[string]interface{}) {
"applicationName": "%v",
"source": "%v",
"model": "%v",
- "type": "prompt"
+ "type": "prompt",
+ "platform": "%v",
+ "generation": "%v",
+ "job.name": "doku"
}
}`,
currentTime,
@@ -219,7 +245,9 @@ func configureNewRelicData(data map[string]interface{}) {
data["endpoint"],
data["applicationName"],
data["sourceLanguage"],
- data["model"]),
+ data["model"],
+ platform,
+ call_type),
}
// Combine the individual log entries into a full JSON payload
@@ -238,8 +266,8 @@ func configureNewRelicData(data map[string]interface{}) {
"type": "gauge",
"value": %v,
"timestamp": %s,
- "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v"}
- }`, data["requestDuration"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["requestDuration"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
}
// Join the individual metric strings into a comma-separated string and enclose in a JSON array.
jsonData := fmt.Sprintf(`[{"metrics": [%s]}]`, strings.Join(jsonMetrics, ","))
@@ -262,9 +290,12 @@ func configureNewRelicData(data map[string]interface{}) {
"source": "%v",
"model": "%v",
"imageSize": "%v",
- "imageQuality": "%v"
+ "imageQuality": "%v",
+ "platform": "%v",
+ "generation": "%v",
+ "job.name": "doku"
}
- }`, data["requestDuration"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["imageSize"], data["imageQuality"]),
+ }`, data["requestDuration"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["imageSize"], data["imageQuality"], platform, call_type),
fmt.Sprintf(`{
"name": "doku.LLM.Usage.Cost",
"type": "gauge",
@@ -277,9 +308,12 @@ func configureNewRelicData(data map[string]interface{}) {
"source": "%v",
"model": "%v",
"imageSize": "%v",
- "imageQuality": "%v"
+ "imageQuality": "%v",
+ "platform": "%v",
+ "generation": "%v",
+ "job.name": "doku"
}
- }`, data["usageCost"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["imageSize"], data["imageQuality"]),
+ }`, data["usageCost"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["imageSize"], data["imageQuality"], platform, call_type),
}
// Join the individual metric strings into a comma-separated string and enclose in a JSON array.
@@ -312,9 +346,12 @@ func configureNewRelicData(data map[string]interface{}) {
"applicationName": "%v",
"source": "%v",
"model": "%v",
- "type": "prompt"
+ "type": "prompt",
+ "platform": "%v",
+ "generation": "%v",
+ "job.name": "doku"
}
- }`, currentTime, promptMessage, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]))
+ }`, currentTime, promptMessage, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type))
}
// Build the image log
@@ -327,9 +364,12 @@ func configureNewRelicData(data map[string]interface{}) {
"applicationName": "%v",
"source": "%v",
"model": "%v",
- "type": "image"
+ "type": "image",
+ "platform": "%v",
+ "generation": "%v",
+ "job.name": "doku"
}
- }`, currentTime, data["image"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]))
+ }`, currentTime, data["image"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type))
// Combine the individual log entries into a full JSON payload
jsonData = fmt.Sprintf(`[{"logs": [%s]}]`, strings.Join(logs, ","))
@@ -346,15 +386,15 @@ func configureNewRelicData(data map[string]interface{}) {
"type": "gauge",
"value": %v,
"timestamp": %s,
- "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "audioVoice": "%v"}
- }`, data["requestDuration"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["audioVoice"]),
+ "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "audioVoice": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["requestDuration"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["audioVoice"], platform, call_type),
fmt.Sprintf(`{
"name": "doku_llm.UsageCost",
"type": "gauge",
"value": %v,
"timestamp": %s,
- "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "audioVoice": "%v"}
- }`, data["usageCost"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["audioVoice"]),
+ "attributes": {"environment": "%v", "endpoint": "%v", "applicationName": "%v", "source": "%v", "model": "%v", "audioVoice": "%v", "platform": "%v", "generation": "%v", "job.name": "doku"}
+ }`, data["usageCost"], currentTime, data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], data["audioVoice"], platform, call_type),
}
// Join the individual metric strings into a comma-separated string and enclose in a JSON array.
@@ -376,9 +416,12 @@ func configureNewRelicData(data map[string]interface{}) {
"applicationName": "%v",
"source": "%v",
"model": "%v",
- "type": "prompt"
+ "type": "prompt",
+ "platform": "%v",
+ "generation": "%v",
+ "job.name": "doku"
}
- }`, currentTime, data["prompt"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"]),
+ }`, currentTime, data["prompt"], data["environment"], data["endpoint"], data["applicationName"], data["sourceLanguage"], data["model"], platform, call_type),
}
// Combine the individual log entries into a full JSON payload
diff --git a/src/ingester/db/db.go b/src/ingester/db/db.go
index af817b0a..cadbb00c 100644
--- a/src/ingester/db/db.go
+++ b/src/ingester/db/db.go
@@ -69,7 +69,6 @@ func PingDB() error {
attempt++
err := db.Ping()
if err == nil {
- log.Info().Msg("Successfully connected to the database")
return nil
}
@@ -223,6 +222,7 @@ func initializeDB() error {
if dbErr != nil {
return
}
+ log.Info().Msg("Successfully connected to the database")
db.SetMaxOpenConns(dbConfig.MaxOpenConns)
db.SetMaxIdleConns(dbConfig.MaxIdleConns)