graphite-dev team mailing list archive

[Anatoliy Dobrosynets <question402437@xxxxxxxxxxxxxxxxxxxxx>]

New question #402437 on Graphite:
https://answers.launchpad.net/graphite/+question/402437

Hi,

need to interpret the graph:  http://imgur.com/a/NCFBS
Graph displays relay-to-cache-destination metrics ( carbon.relays.carbon-relay-a.destinations.carbon-cache-2:a.* )


We noticed many metric values increased 4-6 times, first started on Sept 14. Most of those affected metrics are statsd counters.
Looking into carbon internal metrics we see something changed on that date.

attemptedRelays was different from Sent and  on Sept 14 they became equal. 
At the same date, RelayMaxQueueLength decreased significally.

Does this mean that relay was already unhealthy and it was loosing/dropping/rejecting some datapoints before Sept 14 ?
Does this mean that after Sept 14, it's got healthy and started to deliver all datapoints to destination carbon-cache ?

Carbon-relay receives ~ 220K metrics/min  ,   only ~ 22K of them are relayed to that specific carbon-cache destination.

relay settings:
MAX_QUEUE_SIZE = 1000
USE_FLOW_CONTROL = True

destination carbon settings:
MAX_CACHE_SIZE = inf
USE_FLOW_CONTROL = True


-- 
You received this question notification because your team graphite-dev
is an answer contact for Graphite.