SLA Monitoring¶

Track your uptime promises and delivery targets — know before your customers do if you're falling short.

What SLAs to Track¶

For an email server, three SLAs matter most:

SLA	Target	Measurement
Uptime	99.9% (8.7h downtime/year)	Health monitor availability
Delivery Success	> 95% within 5 minutes	Postfix delivery metrics
API Response Time	p95 < 500ms	FastAPI metrics

Uptime Tracking¶

Uptime is the percentage of time all critical services are available.

What Counts as Downtime¶

Scenario	Counts as Downtime?
Postfix down	Yes
API returning 500s	Yes
Dovecot unreachable	Yes
MySQL down	Yes
Grafana down	No (monitoring, not core)
Scheduled maintenance	Depends on your SLA definition
Single failed health check	No (needs sustained failure)

Prometheus Queries¶

PromQL

# Uptime percentage over 30 days
avg_over_time(up{job="api"}[30d]) * 100

# Per-service uptime
avg_over_time(mailyte_service_up[30d]) * 100

# Combined system uptime (all critical services)
(
  avg_over_time(mailyte_service_up{service="postfix"}[30d])
  * avg_over_time(mailyte_service_up{service="dovecot"}[30d])
  * avg_over_time(mailyte_service_up{service="api"}[30d])
  * avg_over_time(mailyte_service_up{service="mysql"}[30d])
) * 100

# Minutes of downtime in the last 30 days
(1 - avg_over_time(up{job="api"}[30d])) * 30 * 24 * 60

Uptime Table¶

Target	Max Downtime/Month	Max Downtime/Year
99.0%	7.3 hours	3.65 days
99.5%	3.65 hours	1.83 days
99.9%	43.8 minutes	8.77 hours
99.95%	21.9 minutes	4.38 hours
99.99%	4.38 minutes	52.6 minutes

Delivery SLA¶

Track what percentage of emails are delivered successfully within your target window.

PromQL

# Delivery success rate (last 24h)
(
  increase(postfix_delivery_total[24h])
  - increase(postfix_bounce_total[24h])
)
/ increase(postfix_delivery_total[24h]) * 100

# Emails delivered within 5 minutes (percentage)
histogram_quantile(1,
  rate(postfix_delivery_delay_seconds_bucket{le="300"}[1h])
)
/ rate(postfix_delivery_delay_seconds_count[1h]) * 100

# Average delivery time
rate(postfix_delivery_delay_seconds_sum[1h])
/ rate(postfix_delivery_delay_seconds_count[1h])

Delivery SLA targets:

Metric	Target	Alert Threshold
Delivery rate	> 95%	< 93%
Delivered within 1 min	> 80%	< 70%
Delivered within 5 min	> 95%	< 90%
Bounce rate	< 3%	> 5%

API Response Time SLA¶

PromQL

# Percentage of requests under 500ms
sum(rate(http_request_duration_seconds_bucket{le="0.5"}[1h]))
/ sum(rate(http_request_duration_seconds_count[1h])) * 100

# p95 response time trend
histogram_quantile(0.95, rate(http_request_duration_seconds_bucket[5m]))

# Requests meeting SLA by endpoint
sum by (endpoint) (
  rate(http_request_duration_seconds_bucket{le="0.5"}[1h])
)
/ sum by (endpoint) (
  rate(http_request_duration_seconds_count[1h])
) * 100

SLA Dashboard¶

Build a dedicated SLA dashboard in Grafana:

Text Only

+----------------------------+----------------------------+
|    System Uptime (30d)     |    Delivery Rate (30d)     |
|    99.97%  [gauge]         |    96.2%  [gauge]          |
+----------------------------+----------------------------+
|   API SLA Compliance       |   Delivery Time SLA        |
|   98.5% < 500ms [gauge]   |   94.1% < 5min [gauge]     |
+----------------------------+----------------------------+
|   Uptime Over Time (daily, time series)                 |
|   ------------------------------------------------      |
|   Shows daily uptime % with 99.9% target line           |
+---------------------------------------------------------+
|   SLA Breach Log (table)                                |
|   Date | Service | Duration | Impact                    |
+---------------------------------------------------------+

Gauge Thresholds¶

Configure Grafana gauges with color-coded thresholds:

Range	Color	Meaning
> target	Green	Meeting SLA
target - 1% to target	Yellow	Close to breaching
< target - 1%	Red	SLA breached

SLA Alerting Rules¶

YAML

groups:
  - name: sla
    rules:
      - alert: UptimeSLAAtRisk
        expr: avg_over_time(up{job="api"}[24h]) < 0.999
        for: 1h
        labels:
          severity: warning
        annotations:
          summary: "24h uptime dropped below 99.9%"
          description: "Current 24h uptime: {{ $value | humanizePercentage }}"

      - alert: DeliverySLABreach
        expr: >
          (increase(postfix_delivery_total[24h])
           - increase(postfix_bounce_total[24h]))
          / increase(postfix_delivery_total[24h]) < 0.95
        for: 30m
        labels:
          severity: critical
        annotations:
          summary: "Delivery rate below 95% SLA target"

      - alert: APILatencySLABreach
        expr: >
          histogram_quantile(0.95,
            rate(http_request_duration_seconds_bucket[1h])
          ) > 0.5
        for: 15m
        labels:
          severity: warning
        annotations:
          summary: "API p95 latency exceeds 500ms SLA target"

Monthly SLA Reports¶

Generate a monthly report with this script:

Bash

#!/bin/bash
# monthly-sla-report.sh

PROM="http://localhost:9090"
PERIOD="30d"

echo "=== Mailyte SLA Report ==="
echo "Period: Last 30 days"
echo "Generated: $(date)"
echo ""

echo "--- Uptime ---"
curl -s "$PROM/api/v1/query" \
  --data-urlencode "query=avg_over_time(up{job=\"api\"}[$PERIOD]) * 100" \
  | python3 -c "
import json, sys
data = json.load(sys.stdin)
for r in data['data']['result']:
    print(f\"  API Uptime: {float(r['value'][1]):.3f}%\")
"

echo ""
echo "--- Delivery ---"
curl -s "$PROM/api/v1/query" \
  --data-urlencode "query=(increase(postfix_delivery_total[$PERIOD]) - increase(postfix_bounce_total[$PERIOD])) / increase(postfix_delivery_total[$PERIOD]) * 100" \
  | python3 -c "
import json, sys
data = json.load(sys.stdin)
for r in data['data']['result']:
    print(f\"  Delivery Rate: {float(r['value'][1]):.2f}%\")
"

echo ""
echo "--- API Latency ---"
curl -s "$PROM/api/v1/query" \
  --data-urlencode "query=histogram_quantile(0.95, rate(http_request_duration_seconds_bucket[$PERIOD]))" \
  | python3 -c "
import json, sys
data = json.load(sys.stdin)
for r in data['data']['result']:
    ms = float(r['value'][1]) * 1000
    print(f\"  p95 Response Time: {ms:.1f}ms\")
"

Error Budgets¶

An error budget is the inverse of your SLA — it's how much downtime you're "allowed."

For a 99.9% uptime SLA over 30 days:

Total minutes: 43,200
Error budget: 43.2 minutes
Burn rate: How fast you're consuming the budget

PromQL

# Error budget remaining (percentage)
1 - (
  (1 - avg_over_time(up{job="api"}[30d]))
  / (1 - 0.999)
)

# Error budget burn rate (>1 means you'll breach the SLA)
(1 - avg_over_time(up{job="api"}[1h]))
/ (1 - 0.999) * 720

Tip: When your error budget is nearly exhausted, freeze deployments and focus on stability. It's a clear, data-driven signal that reliability work needs to happen now.