In the recent Ores rollout, it was clear that the exporter was [[ https://grafana.wikimedia.org/d/HIRrxQ6mk/ores?orgId=1&from=1575393959648&to=1575409577679 | not relaying all metrics received ]]Around half of the metrics are being relayed.
Even if the last rule is to drop from Prometheus, all metrics should be passed to the relay.`udpreplay` direct to `socat` gets around 26k metrics from a raw packet capture.
```
$ touch direct.txt && socat -t 0 FILE:direct.txt udp-listen:9125,fork
$ ./udpreplay --pps 2000 --host 127.0.0.1 --port 9125 ores1001.pcap
$ cat direct.txt | sed 's/ores.ores1001/\nores.ores1001/g' | sed '/^$/d' | wc -l
26412
```
The current version of statsd_exporter (0.9.0+ds1-1), when put into the middle, gets around 13k metrics.
```
$ touch forwarded_current.txt && socat -t 0 FILE:forwarded_current.txt udp-listen:9125,fork
$ ./statsd_exporter_0.9.0+ds1-1 --statsd.mapping-config=statsd_exporter.conf --statsd.listen-udp=:8125 --statsd.relay-address=127.0.0.1:9125
$ ./udpreplay --pps 500 --host localhost --port 8125 ores1001.pcap
$ cat forwarded_current.txt | sed 's/ores.ores1001/\nores.ores1001/g' | sed '/^$/d' | wc -l
13812
```