Hey all,
I was experimenting with the HTTP Delay Fault Injection. Is there a way to keep track on which timestamps the fault injection is active? I am already scraping Istio metrics into Prometheus (Google Kubernetes Cluster). This is critical for my work. Thank you!
I think you could get this information from the proxy logs generated at the source proxies. They should have the FI flag set. The Prom metrics will not be able to give you exact timestamps, just the count of requests failed with FI. If you use Kiali you can click graph edges and use the Flags tab in the side panel to get a breakdown of requests failing with specific flags. And you can use the Logs tab in Workload Detail to see the proxy logs for a workload. You can get a parsed view of an individual log entry by clicking on it’s associated info icon.
Thank you for replying! Is there a way to scrape the raw data? As I understand these are visual representations of the available data. I wish to use the data for some machine learning models.
To know the times I think you would need to use the proxy logs and identify the applicable log entries via the FI response flag.