timescale / tsbs

Time Series Benchmark Suite, a tool for comparing and evaluating databases for time series data
MIT License
1.24k stars 293 forks source link

tsbs_load_timescale data loss #245

Open GreateFang opened 7 months ago

GreateFang commented 7 months ago

I gennerated a iot dataset and load them to timescaleDB and another database, while the num of data load in timescaleDB less than the another DB.

image

I check the dataset and found some data cannot found in timescale. follow is a example: image image image

GreateFang commented 7 months ago
image

I think the reason is the logic here :If a truck has already reported a tag to table tags once, it will not report again, even if the truck later reports a different tag (simulating imperfect data in the real world).

I think the logic here will also result in will inconsistent tags data when running tsbs_load_timescale multiple times with the same data set. because the tags data is influenced by the order of batch writes.