Metrics
How the search is doing, in counters. A fixed probe search runs on the server every ten minutes; this page shows its time and result count, its last seven days, the engines' own statistics, and whether the search, the Keep and the Drive layer answer.
What is counted: the probe's searches (one fixed query of ours, every ten minutes), the engines' result counts, median response times, reliability and error rates as the instance keeps them, and three health checks. Nothing per person and nothing per query exists: no query, no address and no identifier is counted, stored or sent, here or anywhere behind this page.
The numbers come from one JSON file the probe writes every run; how this search works.
Reading the counters…
Last probe
– ms
– results
Seven days
– of – runs ok
median – ms · – results
Health
- Search
- Keep
- Drive (Photos too)
Probe time, seven days
Each point is one probe run that returned results (milliseconds); the dashed line is the median.
–
Engines
From the probe: which engines answered its last search, and which failed. Below them, the instance's own statistics since it last started (results returned, median response time, reliability, the share of requests that failed), once it keeps them.
–
| Engine | Results | Median | Reliability | Errors |
|---|