Same behavior on all 1.3 and 1.4.
On 1.4 we have simple info about DB cardinality:
cardinality estimation ---------------------- 1856641
taking show series on this DB
show series limit 1 offset 0
shows 1 series as we expect but after almost 30 seconds !!!
real 0m29.573s user 0m0.009s sys 0m0.007s
> show series limit 100 offset 10 real 0m31.508s user 0m0.000s sys 0m0.010s
show series limit 10000 offset 10 real 0m35.080s user 0m0.148s sys 0m0.063s
Always more than 30 second
We have other DB with more tags and it takes even 7 minutes to get 1 series from a schema which is totally inefficient.
It is no different how big limit we take, it is always building whole series list on the backend?
Is there any way to improve metadata queries in influxdb, cache, some configuration improvements?