How does the retention on Graphite storage schema work?

[亡魂溺海] 提交于 2021-02-05 08:01:05

问题


I am trying to understand how the retentions in the storage-schema.conf of Graphite work.

This is my retention configuration.

[prod_retention]
pattern = prod\..*
retentions = 10s:30d,1m:90d

Does this mean the data retention is for 120 days (30d+90d) or for just 90 days ?


回答1:


That configuration means that a rolling window of 30d will have metrics at a 10s resolution and after data points exceed 30d of age, they are downsampled to a rolling window of 90d that will have metrics stored at a 1m resolution. These resolutions are stored in the same data file. Your longest data retention in this file would be 90d.

The aggregation function that will be used for this file when rolling up 10s data points to the 1m resolution is defined within storage-aggregations.conf (using average/mean by default).

See configuration detailed documented here: http://graphite.readthedocs.org/en/latest/config-carbon.html#storage-schemas-conf and http://graphite.readthedocs.org/en/latest/config-carbon.html#storage-aggregation-conf



来源:https://stackoverflow.com/questions/29558912/how-does-the-retention-on-graphite-storage-schema-work

标签
易学教程内所有资源均来自网络或用户发布的内容,如有违反法律规定的内容欢迎反馈
该文章没有解决你所遇到的问题?点击提问,说说你的问题,让更多的人一起探讨吧!