CY22 H2 Goals:
- Remove extra hop and reduce request latency by enabling IPFS gateway to lookup Indexer directly with reframe (without having to go through hydra boosters)
- Increase Index coverage (esp. for ipfs gateway traffic) by ingesting index from IPFS nodes through reframe publish endpoint
Current State (Stage 0)
Hydra uses reframe to lookup indexer ✅
Done state:
- Hydras in production across the whole fleet hit STI using the reframe endpoint
- Hydras in production across the whole no longer have/use the old STI findProviders API
- Hydras have clean metrics metrics for:
- Number of calls to STI
- Number of calls that succeeded with a 2xx (success) response (regardless if STI has providers for the given CID or not)
- Number of calls that fataled (e.g., 5xx due to server issue)
- For each 2xx responses, record of the number of providers in the response. This allows us to say the the p90 of responses have X providers.
Tracking issue: https://github.com/libp2p/hydra-booster/issues/162
ToDo/Issues:
- [x] Roll out reframe fully to indexer and sunsetting existing API
- [x] Update with newer version of caching with reframe
- [x] Metrics to show reframe call success / fail
- [x] Address timeout issue (for requests not cached)
- [x] Fix high error rate: Dashboard with status metric ‘reframe server returned error’ vs ‘success’ showing a high 15% error rate on the reframe requests
DRIs:
- Roll out reframe & hydra deployment: Petar (Gus might help)