fix(reset_budget_job): write minimal reset updates

[pyguy]

Summary

• Fixes reset budget key/team writes by sending only identifier, spend, and budget_reset_at through update_data.
• Adds dict payload support to PrismaClient.update_data update_many for key/team, preserving existing object payload behavior.
• Adds regression coverage for key/team rows with object_permission_id and budget_limits, including dict/object update_data payload variants.

Fixes #27730.

Testing

• uv run pytest tests/test_litellm/proxy/common_utils/test_reset_budget_job.py -x -vv (49 passed)
• uv run pytest tests/litellm_utils_tests/test_proxy_budget_reset.py -x -vv (14 passed)
• Live k8s dev validation on ai-tools-dev/litellm: unpatched pod reproduced the Prisma DataError for key and team reset rows with object_permission_id + budget_limits; patched pod reset both rows successfully with spend=0.0, advanced budget_reset_at, retained object_permission_id and budget_limits, and failure hook count 0. Both running dev pods were hot-patched with the implementation files afterward.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[greptile-apps]

Greptile Summary

This PR fixes a Prisma DataError that occurred when resetting budgets for keys and teams that had object_permission_id or budget_limits populated — those fields were being forwarded wholesale to the DB update and rejected by Prisma. The fix narrows the write payload to only the three fields that should change: token/team_id, spend, and budget_reset_at.

• reset_budget_job.py: Projects updated_keys / updated_teams into minimal dicts before passing to update_data, eliminating the stray field problem at the call site.
• utils.py: Adds a dict-payload branch in the update_many handlers for both the key and team tables so the new minimal dicts are routed correctly alongside the existing object-based path.
• Tests: Fixtures gain token/team_id fields; assertions tighten to check the exact set of keys written, and four new unit tests cover both dict and object payload variants through the updated update_data code.

Confidence Score: 4/5

Safe to merge — the change is narrowly scoped to the budget reset write path, the root-cause fix is correct, and live k8s validation plus 63 passing tests back it up.

The core fix is sound: extracting a minimal dict before the DB write prevents Prisma from receiving unexpected JSON/relation fields. One cosmetic issue exists in utils.py — data_json["token"] = token is assigned redundantly after jsonify_object already includes it from the dict comprehension — but this has no effect on runtime behaviour. No logic regressions were identified.

litellm/proxy/utils.py — the redundant token reassignment on line 3885 is worth cleaning up, but it has no runtime impact.

Important Files Changed

Filename	Overview
litellm/proxy/common_utils/reset_budget_job.py	Projects only token/team_id, spend, and budget_reset_at into minimal dicts before calling update_data, fixing the Prisma DataError caused by unexpected fields like object_permission_id and budget_limits.
litellm/proxy/utils.py	Adds dict-payload branch to update_many for key and team tables; contains a redundant data_json["token"] = token assignment after the token is already present from the dict comprehension passed to jsonify_object.
tests/test_litellm/proxy/common_utils/test_reset_budget_job.py	Adds token/team_id fields to test fixtures, updates assertions to reflect dict-payload shape, and adds four new direct unit tests for the update_data dict and object paths for both key and team tables.
tests/litellm_utils_tests/test_proxy_budget_reset.py	Test fixtures updated to include token/team_id fields; assertions changed from id-based checks to token/team_id with key-set assertions that verify no extra fields leak into the DB payload.

_{Reviews (1): Last reviewed commit: "test(reset_budget_job): cover update_dat..." | Re-trigger Greptile}

[oss-pr-review-agent-shin]

🤖 litellm-agent: This PR is currently BLOCKED from merge.

Score: 3/5 ❌

Why blocked:

• 1 PR-related CI failure (Size gate: tests (+286) exceed code (+75) by more than 3× — over-specified or feature too thin. Add the oversized-ok label if intentional.) (pr_related_failures, -2 pts)

Details: Score docked for: 1 PR-related CI failure (Size gate: tests (+286) exceed code (+75) by more than 3× — over-specified or feature too thin. Add the oversized-ok label if intentional.).

Fix the issues above and push an update — the bot will re-review automatically.

Note: This bot is still in beta and might not always work as expected. Please share any feedback via Slack.

[pyguy]

All previously reported blockers are addressed on the latest commit: the size gate is within bounds (code +74, tests +222), Codecov patch coverage passes, and all GitHub checks are green. Requesting a fresh LiteLLM agent review.

[pyguy]

@krrish-berri any chance to have this merged? thanks.

[pyguy]

Fixed by #29358.