Posted on April 27, 2020 2:12 pm
 |  Asked by Raymond Driedonks
Print Friendly, PDF & Email

Hi, just read “Streaming EOS telemetry states to Prometheus” by Tamas. Great article. Does anyone know if, and if so, how the push benefits of streaming telemetry are preserved while using pull based Prometheus? Will all SysDB state changes be available in Prometheus with only a delay defined by the scrape interval? How about the timestamps?


Posted by Tamas Plugor
Answered on April 27, 2020 2:26 pm

Hi Raymond!

Thank you, glad you liked it!

Yes, since Prometheus is polling data from ocprometheus which is subscribed to the metrics you defined in the yaml file (via TerminAttr or gNMI API(another article will talk about gNMI way soonish)),  and keeps the states locally, but always keeps the latest value in mind, so essentially, you will get the state from Sysdb/Smash at the time of scraping, so you won't have data points between (t) and (t-delta). Timestamps are added by Prometheus itself when the data gets in.



Hi Tamas, Thanks. So in case of multiple changes for a given metric within a scrape interval, only the most recent one is added to Prometheus and granularity depends on the scrape interval.
(Raymond Driedonks at April 28, 2020 9:06 am)

Post your Answer

You must be logged in to post an answer.