Ikibazo gihuza ibibazo · Hagati Urwego
Gukurikirana ML hamwe na Prometheus na Grafana Ikibazo
Hitamo ubwoko bwa metero ya Prometheus hamwe na Grafana kubitekerezo bya serivisi zifatika, mugihe wirinze ubukaridinali nibibazo byibanga.
Ikibazo 1 Bya 8
Nigute Prometheus ikusanya metero muri serivisi ntangarugero?
Reba buri kibazo8 ibibazo
Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.
- Nigute Prometheus ikusanya metero muri serivisi ntangarugero?
- Scrape a metrics endpoint exposed by the service
- Read every user's prediction database directly
- Retrain the model at each interval
- Inspect dashboard screenshots
- Ni ubuhe bwoko bwa metric bujyanye no kubara ibyifuzo?
- Gauge
- Counter
- Histogram bucket label only
- Dashboard annotation
- Kuki wirinda indangamuntu nkikirango cya Prometheus?
- It prevents latency measurement
- It changes the model weights
- It can create an unbounded number of time series
- It makes the metric name invalid
- Ni iki amateka yubukererwe ashobora gutanga?
- A direct causal explanation
- A guarantee of model accuracy
- A raw prompt transcript
- Aggregatable bucket counts for latency distributions
- Ni ubuhe butumwa bwa Grafana bwerekana kuri serivisi zifatika?
- Only model training accuracy
- Service latency, errors, traffic and relevant model signals
- Every raw user prompt
- Only the most recent prediction
- Why separate service health alerts from model-quality monitoring?
- Quality never changes after deployment
- Prometheus cannot store model metrics
- A service can be available while predictions are poor, or vice versa
- Latency is the only useful metric
- Where should raw request IDs or sensitive prompt text generally be handled?
- Public dashboard titles
- Unbounded Prometheus labels
- Metric names
- Controlled logs or traces with access and retention policies
- What does a pending period in an alert rule help with?
- Avoiding alerts on brief transient metric spikes
- Reducing metric cardinality
- Improving model calibration
- Changing histogram buckets after collection