Closed spoofedpacket closed 5 years ago
carbon-clickhouse uses disk queue for batch upload to clickhouse. The most useful metric - carbon.agents.*.upload.*.unhandled
. This is lag in seconds between receiving a point and when it starts to appear on the graphs, if you use the default chunk-interval="1s"
Brilliant, thanks!
Does carbon-clickhouse have any equivalent to the internal carbon-cache PPU/update operations/update time metrics? Or is this something that's better exposed by clickhouse itself?
Context: I'm evaluating carbon-clickhouse in comparison to traditional python-carbon + whisper. It appears to be accepting a very high volume of metrics without issue but I'm interested in how it's performing under the hood.