mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-15 04:19:08 +00:00
MCP moves to two large nodes with two workers each, the same two cores per worker it passes with on xlarge. The census jobs only need the shard verifier and coverage, so they install those six locked packages with hashes on a small Docker executor instead of the full test setup. Postgres for the integration contracts allows 300 connections, since every isolated worker opens its own proxy pools. Management, accounting, providers and database keep their current sizes because a run on large machines pushed tests that already sit near their time budgets past them Co-authored-by: Yuneng Jiang <yuneng@berri.ai> |
||
|---|---|---|
| .. | ||
| scripts | ||
| config.yml | ||