In integration tests, at the very beginning, we set enable_seqscan to off. What will force the scheduler to avoid seq scan, if it has such an opportunity (there is an index)
At the end of the test, we look at the statistics of accesses to the table (for each table, postgres counts the number of seq scans and accesses through indexes). If the number of seq scans is greater than 0, it means that there is not enough index for some SELECT
To check, there is a diagnosis of "Tables with missing indexes"
It's like an alternative https://habr.com/ru/companies/tinkoff/articles/454066/
In integration tests, at the very beginning, we set
enable_seqscantooff. What will force the scheduler to avoid seq scan, if it has such an opportunity (there is an index)At the end of the test, we look at the statistics of accesses to the table (for each table, postgres counts the number of seq scans and accesses through indexes). If the number of seq scans is greater than 0, it means that there is not enough index for some SELECT
To check, there is a diagnosis of "Tables with missing indexes"
It's like an alternative https://habr.com/ru/companies/tinkoff/articles/454066/