Page MenuHomePhabricator

diamond network collector loss not accurate
Closed, ResolvedPublic

Description

as discovered in T89846 there is signifcant udp packet loss on graphite1001 but no loss reported in the relevant dashboard at http://gdash.wikimedia.org/dashboards/graphite/ summarizing diamond's tx_drop and rx_drop network metrics

Event Timeline

fgiunchedi claimed this task.
fgiunchedi raised the priority of this task from to High.
fgiunchedi updated the task description. (Show Details)
fgiunchedi subscribed.

the reason for this is that the network collector is separate from ip/tcp/udp collectors which would have been the right thing in this case, since enabling those collector will push a significant amount of metrics we have to figure out statsd first in T90111

no reason to enable udp collector across the board, we can selectively enable it for machines with udp services (likewise tcp)