Code School of Analytics
IntermediateLists & Dictionaries25 points

Which microservices are on fire?

The on-call engineer of a Berlin fintech gets a dump of log lines and wants to know which services have an unhealthy error rate. Each line looks like 2026-03-14 09:15:02 [ERROR] checkout: Payment gateway timeout: a date, a time, the level in square brackets, the service name, a colon, and a free-text message.

  • The level is the first bracketed word of the line and may be written in any case. Levels ERROR and FATAL count as errors; everything else (INFO, WARN, DEBUG) does not.
  • The service name is the text between ] and the first colon after it. Names are case-sensitive. The message may contain colons or more brackets, which must be ignored.

Task: line 1 is an integer threshold T (percent). Line 2 is the number of log lines N. Then come N log lines. A service is unhealthy when errors * 100 >= T * total (total = all its lines). Print each unhealthy service as service: E/total (R%) where R is E*100/total with one decimal. Sort by rate descending, services with exactly equal rates by name ascending. If no service is unhealthy print All services healthy.

Sample tests

Input

30
8
2026-03-14 09:15:02 [ERROR] checkout: Payment gateway timeout
2026-03-14 09:15:05 [INFO] checkout: Order 4411 created
2026-03-14 09:16:00 [info] search: Query ok
2026-03-14 09:16:10 [WARN] search: slow response: 2300ms
2026-03-14 09:17:00 [Error] auth: Token store down
2026-03-14 09:17:01 [FATAL] auth: Cannot reach redis [retrying]
2026-03-14 09:18:00 [INFO] auth: Login ok
2026-03-14 09:19:00 [DEBUG] checkout: cache warm

Expected output

auth: 2/3 (66.7%)
checkout: 1/3 (33.3%)

Input

90
2
2026-01-01 00:00:01 [INFO] api: start
2026-01-01 00:00:02 [ERROR] api: boom

Expected output

All services healthy

Your answer is also checked against 3 hidden tests.

Need a hint?

Fractions like 1/3 and 2/6 are exactly equal when you compare them as numbers, so they are ties.