feat(observability): add prometheus metrics #37
Conversation
|
It seems like you've provided a comprehensive summary of changes and a delightful poem! Is there anything else you'd like to add or modify? TipsChat with CodeRabbit Bot (
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Review Status
Actionable comments generated: 1
Configuration used: CodeRabbit UI
Files selected for processing (7)
- cmd/feeder/main.go (2 hunks)
- feeder/priceposter/client.go (2 hunks)
- feeder/priceprovider/priceprovider.go (1 hunks)
- feeder/priceprovider/sources/tick_source.go (4 hunks)
- feeder/priceprovider/sources/tick_source_test.go (3 hunks)
- metrics/README.md (1 hunks)
- metrics/metrics.go (1 hunks)
Files skipped from review due to trivial changes (1)
- metrics/README.md
Additional comments: 11
feeder/priceposter/client.go (2)
7-13:
The import of themetricspackage is correctly placed and follows the convention of grouping third-party imports together.122-122:
The addition of the metric increment operation is correctly implemented and follows the logic described in the summary.feeder/priceprovider/priceprovider.go (1)
- 39-51:
The changes to the
NewPriceProviderfunction correctly pass the source type as an additional argument to theNewTickSourcefunction for each case. This aligns with the implementation of observability features described in the summary.feeder/priceprovider/sources/tick_source.go (4)
4-10: The import of the
metricspackage is correctly added to support the new Prometheus metrics functionality.4-22: > Note: This review was outside the patches, so it was mapped to the patch with the greatest overlap. Original lines [19-28]
The
NewTickSourcefunction has been correctly updated to accept a newsourceparameter and initialize thesourcefield in theTickSourcestruct.
43-47: The
sourcefield has been correctly added to theTickSourcestruct to store the source identifier for the metrics.60-67: The metric recording for
PriceSourceLatencyis correctly implemented within theTickSourceloop, capturing the duration of thefetchPricescall and observing it with the appropriate label.feeder/priceprovider/sources/tick_source_test.go (3)
29-32: The test case for
NewTickSourcecorrectly passes the "test_source" string identifier as per the changes described in the summary. This ensures that the test cases are aligned with the updated function signature.64-66: The test case for
NewTickSourcewithin the "price update dropped due to shutdown" scenario has been updated to include the "test_source" string identifier. This maintains consistency with the changes made to theNewTickSourcefunction.83-85: The test case for
NewTickSourcewithin the "logs on price update errors" scenario has been updated to include the "test_source" string identifier. This maintains consistency with the changes made to theNewTickSourcefunction.metrics/metrics.go (1)
- 1-17:
The implementation of Prometheus metrics using the
promautopackage is correct, and the metrics are well-defined with appropriate names and help descriptions. The use of default buckets for the histogram is acceptable for initial implementation but may need to be revisited for fine-tuning based on actual latency distributions observed.
cmd/feeder/main.go
Outdated
| http.Handle("/metrics", promhttp.Handler()) | ||
| http.ListenAndServe(":3000", nil) |
There was a problem hiding this comment.
The code to start the HTTP server for Prometheus metrics is not wrapped in a goroutine, which means that the ListenAndServe call will block the main function from proceeding. This could prevent the handleInterrupt function from working as intended since it's supposed to listen for an interrupt signal to gracefully shut down the feeder.
To fix this, the HTTP server should be started in a separate goroutine to allow the main program to continue running and be able to handle interrupts.
- http.ListenAndServe(":3000", nil)
+ go func() {
+ if err := http.ListenAndServe(":3000", nil); err != nil {
+ logger.Error().Err(err).Msg("Metrics HTTP server failed")
+ }
+ }()Commitable suggestion
❗ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation.
| http.Handle("/metrics", promhttp.Handler()) | |
| http.ListenAndServe(":3000", nil) | |
| http.Handle("/metrics", promhttp.Handler()) | |
| go func() { | |
| if err := http.ListenAndServe(":3000", nil); err != nil { | |
| logger.Error().Err(err).Msg("Metrics HTTP server failed") | |
| } | |
| }() |
Wrap the `http.ListenAndServe` in a goroutine to prevent conflicts with the interrupt handler
There was a problem hiding this comment.
Review Status
Actionable comments generated: 1
Configuration used: CodeRabbit UI
Files selected for processing (1)
- cmd/feeder/main.go (2 hunks)
Additional comments: 2
cmd/feeder/main.go (2)
5-10:
The import statements for the HTTP server functionality and Prometheus client are correctly added.61-61:
Clarify the purpose of theselect{}statement blocking indefinitely, as it may not be obvious to maintainers why this is needed.
| http.Handle("/metrics", promhttp.Handler()) | ||
| go func() { | ||
| if err := http.ListenAndServe(":3000", nil); err != nil { | ||
| logger.Error().Err(err).Msg("Metrics HTTP server failed") | ||
| } | ||
| }() |
There was a problem hiding this comment.
Consider logging a message when the HTTP server for metrics starts successfully to provide clear feedback on the server status.
This PR defines two metrics that are used to monitor the performance and behavior of the pricefeeder application.
pricefeeder_voting_periods_total
This is a counter metric that keeps track of the total number of voting periods the pricefeeder has participated in. A voting period is a specific duration during which votes are collected. This metric is incremented each time the pricefeeder participates in a voting period.
pricefeeder_price_source_latency_seconds
This is a histogram metric that measures the latency of querying a price source in seconds. The latency is the amount of time it takes for the pricefeeder to receive a response after it has sent a request to the price source. This metric is useful for monitoring the performance of the price sources and identifying any potential issues or delays.
The source label is used to differentiate the latencies of different price sources.
Summary by CodeRabbit
New Features
/metricsendpoint for Prometheus scraping.Enhancements
SendPricesfunction to increment a counter for tracking voting periods.NewTickSourcefunction.Documentation
Tests
NewTickSourcefunction calls.