Better error handling when collecting CPU temps

Log why we couldn't collect the temperature, and set metric to NaN if
the CPU should support temperature collection but had an error.

collector/cpu_freebsd.go

@@ -17,10 +17,12 @@ package collector
 
 import (
 	"fmt"
+	"math"
 	"strconv"
 	"unsafe"
 
 	"github.com/prometheus/client_golang/prometheus"
+	"github.com/prometheus/common/log"
 	"golang.org/x/sys/unix"
 )
 
@@ -130,10 +132,18 @@ func (c *statCollector) Update(ch chan<- prometheus.Metric) (err error) {
 		ch <- c.cpu.mustNewConstMetric(float64(t.idle), lcpu, "idle")
 
 		temp, err := unix.SysctlUint32(fmt.Sprintf("dev.cpu.%d.temperature", cpu))
-		if err == nil {
-			ftemp := float64(temp-2732) / 10
-			ch <- c.temp.mustNewConstMetric(ftemp, lcpu)
+		if err != nil {
+			if err == unix.ENOENT {
+				// No temperature information for this CPU
+				log.Debugf("no temperature information for CPU %d", cpu)
+			} else {
+				// Unexpected error
+				ch <- c.temp.mustNewConstMetric(math.NaN(), lcpu)
+				log.Errorf("failed to query CPU temperature for CPU %d: %s", cpu, err)
+			}
+			continue
 		}
+		ch <- c.temp.mustNewConstMetric(float64(temp-2732)/10, lcpu)
 	}
 	return err
 }