Closed finestructure closed 1 week ago
Putting this back in draft, because I'm seeing a huge performance regression when running the tests locally with the postgres:16.3-alpine
container: 360 seconds vs 90 seconds.
I'm guessing it's related to the temp fs flag --tmpfs /pgdata:rw,noexec,nosuid,size=1024m
we're using. Needs investigation.
No, it's Postgres itself. Here are the timing just running the AnalyzerTests
:
❯ make db-down-test db-up-test && sleep 3 && time swift test --disable-automatic-resolution --filter AppTests.AnalyzerTests
for different versions of Postgres (plus with/without tempfs in two cases):
Next guess it's related to the table truncation we're doing in db test setup.
Both
dev
andprod
are on version 16.3 now.In order to switch over locally, run
⚠️ This will delete local data ⚠️
Check the postgres version via the following SQL query: