Error Rate

Error rate is the share of requests or operations that fail during a time window, expressed as a percentage: failed requests divided by total requests. For web services, "failed" most often means an HTTP 5xx response, a timeout or a dropped connection.

Why it matters

Error rate is one of the four golden signals and the most direct measure of whether users are being served. It is the natural basis for an SLI, and a target such as "99.9% of requests succeed" becomes an SLO and an error budget.

Percentages beat raw counts. Fifty errors per minute is a crisis at 100 requests per minute and noise at 5 million. Rates also scale with traffic, so one alert threshold keeps working through growth.

Example

requests in window:  120,000
5xx responses:           360
error rate = 360 / 120,000 = 0.3%

Against a 99.9% success target, the allowed error rate is 0.1%, so this window burned its budget at roughly three times the allowed pace.

Common mistakes

• Counting 4xx responses as server errors. A flood of 404s from a crawler is not your outage, though a sudden 401 spike after a deploy can be.
• Ignoring low-traffic periods, where one failure at 3 a.m. is a huge percentage. Add a minimum request count to the alert.
• Measuring at the application only and missing errors the load balancer returned before the app saw the request.

Related

• Golden signals
• SLI
• Error budget
• HTTP status code

यह पेज सामान्य जानकारी है, जिसे हमारी संपादकीय टीम ने लिखा है और दस्तावेज़ीकरण बदलने पर अपडेट किया जाता है। सर्वर और प्रदाता अलग-अलग होते हैं, इसलिए बदलाव पहले सुरक्षित तरीके से जाँचें। देखें: संपादकीय नीति और अस्वीकरण.