Split datasource across multiple tiers

Hi druidfolks,

In our company we want to utilize different tiers of historicals. My major question is, whether it is possible to partition one datasource, such that say, the latest 3 days(1-3) are in tier A and next 27 days (4-30) are in tier B (without having the first 3 days in B as well)

I thought this should be possible, and setup these rules (see image):

For some reason, the data for day 4 is still in the first tier, filling it up, and not in tier B. What am I missing ?

The blue outlined segment is for day 4, but if I check which machine it’s one, I see it’s still loaded on machines (?) of Tier A

  • On a sidenote, I just wonder why it even shows 3-5 machines a segment is on, even though clearly the rule defines only 1 replica of a segment (red A) - ??

  • Do note, that I also tried without the (purple) Drop-P3D rule, just having 2 Load and 1 DropForever-rule.

Thanks in advance on how to define rules for this.

Cheers,

André

Hi Andre,

I have a similar use-case. Did you figure out how to configure the load rules for your use-case?

Thanks,

Avinash

have similar issue

在 2018年4月9日星期一 UTC+8下午5:09:58,an…@simplaex.com写道: