Currently running grafana v5.0.4 with graphite 0.9.13 both on AWS. Our infrastructure is on AWS and highly dynamic. Instances are spun up and down very often which I believe is causing graphs to break when you try to go back over 24 hours. From testing, I believe because we use wildcards in our graph configuration, it looks up both running and nonrunning instances.
We cant delete instance information of nonrunning instances because they are needed for historical purposes. Has someone experienced this issue or have any workaround?
On the 24-hour chart you see data plotted for 4 items: batch, nlp, web, jboss, batch2, pickup while on the past 7 days chart that includes the same period, you only see the data plotted for web, jboss and batch2.
the data for the other types (batch, nlp) are not rendering.
I speculate that this might be because batch and nlp hosts have been destroyed and redeployed during this 7-day period, while the others were not. The break you see was when carbon-cache was restarted just in case someone was wondering. Thanks