I found that countDocuments() take different time in different copy of the same db. The case is next:
- counts documents same (about 70k)
- structure documents the same
- instance the same 4vcpu і 32 ram.
Slower db stats:
Faster db stats:
As you can see slower db has bigger "avgObjSize" more than 10 times. And "countDocuments({filter})" with the same filter runs 10 times longer.
Еhere is some dependence on the avgObjSize(storageSize) and counting documents? Or how I can speed up counting?