Lead
IIS żyje – i to często pod SQL-owym backendem. Pokażę krótki pipeline metryk i logów do Grafany.
1
2
3
4
5
| [[inputs.win_perf_counters]]
[[inputs.win_perf_counters.object]]
ObjectName = "Web Service"
Counters = ["Current Connections","Get Requests/sec","Post Requests/sec"]
Instances = ["_Total"]
|
Telegraf – logi IIS (tail + grok)
1
2
3
4
5
| [[inputs.tail]]
files = ["C:/inetpub/logs/LogFiles/W3SVC1/u_ex*.log"]
from_beginning = false
data_format = "grok"
grok_patterns = ['%{TIMESTAMP_ISO8601:time} %{IPORHOST:client} %{WORD:method} %{URIPATHPARAM:uri} %{NUMBER:status:int} %{NUMBER:sc_substatus:int} %{NUMBER:sc_win32_status:int} %{NUMBER:time_taken:int}']
|
W Grafanie
- Panel
HTTP 5xx, Average time_taken, Requests/sec. - Alert:
5xx rate > 1% przez 5m.
Pro tip
Zsynchronizuj zegary (NTP) — inaczej korelacja metryk/logów to horror.
