fix: retry on download errors

[carlossg]

to avoid failing because of issues like connection reset

[yannh]

Good idea, though this never gives up and doesnt back off incrementallly.. Any library we might be able to use maybe? 🤔

[carlossg]

It just tries once, one second later, but I've used now hashicorp/go-retryablehttp.
If this looks good the tests need to use a real local server to test the retries

[carlossg]

@yannh added a test with a real http server to test the retries

test output

2024/07/26 13:34:55 [DEBUG] GET http://localhost:9163/simulate-reset
2024/07/26 13:34:55 [ERR] GET http://localhost:9163/simulate-reset request failed: Get "http://localhost:9163/simulate-reset": EOF
2024/07/26 13:34:55 [DEBUG] GET http://localhost:9163/simulate-reset: retrying in 1s (2 left)
2024/07/26 13:34:56 using schema found at http://localhost:9163/simulate-reset
2024/07/26 13:34:56 [DEBUG] GET http://localhost:9163/404
2024/07/26 13:34:56 could not find schema at http://localhost:9163/404
2024/07/26 13:34:56 [DEBUG] GET http://localhost:9163/500
2024/07/26 13:34:56 [DEBUG] GET http://localhost:9163/500 (status: 500): retrying in 1s (2 left)
2024/07/26 13:34:57 [DEBUG] GET http://localhost:9163/500 (status: 500): retrying in 2s (1 left)
2024/07/26 13:34:59 failed downloading schema at http://localhost:9163/500: Get "http://localhost:9163/500": GET http://localhost:9163/500 giving up after 3 attempt(s)
2024/07/26 13:34:59 [DEBUG] GET http://localhost:9163/503
2024/07/26 13:34:59 [DEBUG] GET http://localhost:9163/503 (status: 503): retrying in 1s (2 left)
2024/07/26 13:35:00 using schema found at http://localhost:9163/503
2024/07/26 13:35:00 [DEBUG] GET http://localhost:9163
2024/07/26 13:35:00 using schema found at http://localhost:9163

[yannh]

Not sure about creating a cache folder here... Esp as we test the HTTP client?

[yannh]

Hi @carlossg , thanks a lot for the contribution. Your tests are quite a bit more complex than what we had before, but they look good. I iterated on them a little bit, can you let me know what you think? Thanks!

[carlossg]

much better thanks

[yannh]

Acceptance tests are failing, not sure why - they run locally just fine. Unsure if related to this PR 🤔 it ll take me some time to troubleshoot.

[yannh]

I reverted it, it seems to print debug log in some cases. I ll review this later today/this week 🙇