while investigating T101141: UDP rcvbuferrors and inerrors on graphite hosts it occurred to me that diamond (and thus python-statsd) should also be batching statsd metrics, ATM it doesn't (strace from statsdlb). see also upstream issue for multiple udp batches, https://github.com/jsocol/pystatsd/issues/67
recvfrom(3, "servers.ms-fe3001.network.eth3.rx_byte:0.0|g", 16777215, 0, NULL, NULL) = 44 recvfrom(3, "servers.ms-fe3001.network.eth3.rx_fifo:0.0|g", 16777215, 0, NULL, NULL) = 44 recvfrom(3, "servers.ms-fe3001.network.eth3.tx_bit:0.0|g", 16777215, 0, NULL, NULL) = 43 recvfrom(3, "servers.ms-fe3001.network.eth2.rx_compressed:0.0|g", 16777215, 0, NULL, NULL) = 50 recvfrom(3, "servers.ms-fe3001.network.eth3.tx_packets:0.0|g", 16777215, 0, NULL, NULL) = 47 recvfrom(3, "servers.ms-fe3001.network.eth0.rx_fifo:0.0|g", 16777215, 0, NULL, NULL) = 44 recvfrom(3, "servers.ms-fe3001.network.eth1.rx_byte:0.0|g", 16777215, 0, NULL, NULL) = 44 recvfrom(3, "logstash.rate.mediawiki.DBPerformance.INFO:1|c", 16777215, 0, NULL, NULL) = 46 recvfrom(3, "servers.ms-fe3001.network.eth1.rx_frame:0.0|g", 16777215, 0, NULL, NULL) = 45 recvfrom(3, "servers.ms-fe3001.network.eth1.tx_multicast:0.0|g", 16777215, 0, NULL, NULL) = 49 recvfrom(3, "servers.ms-fe3001.network.eth0.rx_packets:19.7166666667|g", 16777215, 0, NULL, NULL) = 57 recvfrom(3, "restbase.sys_parsoid_html_-title--_revision--_tid-.GET.ALL:23|ms", 16777215, 0, NULL, NULL) = 64 recvfrom(3, "servers.ms-fe3001.network.eth0.tx_compressed:0.0|g", 16777215, 0, NULL, NULL) = 50 recvfrom(3, "restbase.sys_page-revisions_page_-title--_revision-.GET.ALL:2|ms", 16777215, 0, NULL, NULL) = 64 recvfrom(3, "servers.ms-fe3001.network.eth0.tx_packets:9.68333333333|g", 16777215, 0, NULL, NULL) = 57