Scripts Run Order — Post-Migration Reference

Quick reference for which scripts to run (and in what order) after migrating fresh data from production.

Prerequisites

• script/.env must have the correct MONGO_CONNECTION_STRING and DATABASE_NAME for the target environment
• Update script/config.json → "scriptRunArray" with the script name before each run
• Run with yarn script-dev (requires Node >=22)

Environment Config

Env	DATABASE_NAME	Connection String Host
Local	dtcea-mumbai-demo	energyaudit.0l6duna.mongodb.net
Dev	dtcea-mumbai-demo-dev	dev.e45p6au.mongodb.net
Stg	dtcea-mumbai-demo-stg	stg.ivmjjjb.mongodb.net

---

Run Order

Scripts have dependencies — run in this exact order.

#	Script Name	Collection	What It Does	Idempotent?	Est. Time
1	seed_aiAuditTestData	transformer, installation, user	Seeds demo data for section S01 (15 TCs, ~135 installations × 7 months). Run via API: POST /seed/ai-audit-test-data with { "clearExisting": true }	Yes	~5s
2	transformer_backfillAuditStatus	transformer	Sets auditStatus = AUDITED or AUDIT_FAILED on all TCs based on lossPercentage presence	Yes (only touches auditStatus: null)	~10s
3	transformer_fixInvalidAuditedTCs	transformer	Fixes TCs incorrectly marked AUDITED (have lossPercentage but missing tcConsumption or installationConsumption) → sets to AUDIT_FAILED	Yes	~10s
4	transformer_fixZeroTcConsumptionLoss	transformer	Fixes migrated TCs with tcConsumption: 0 but lossPercentage: 100 → nulls out loss %, sets AUDIT_FAILED, fixes cascading previousMonthLossPercentage / previousToPreviousMonthLossPercentage	Yes	~30s
5	installation_backfillBillingFlags	installation	Backfills billing boolean flags (unbilled, mnr, vacant, doorLock, zeroConsumption) and numeric billing fields (consumption, demand, collection, emin, emax, billCancellation) by reading production DB's audit[].billing data. Requires productionConnectionString in script input. Excludes section S01 by default.	Yes (overwrites)	10–30 min
6	installation_backfillAvgConsumption	installation	Computes 6-month rolling avg consumption for remarked installations (uses server-side $out + $lookup + $merge)	Yes	5–15 min
7	transformer_backfillHealthScore	transformer	Computes healthScore + healthGrade on all TCs using audit fields, billing counts, and loss data	Yes (overwrites)	3–8 min
8	auditSummary_computeAuditSummary	audit-summaries	MUST BE LAST. Deletes all existing summaries, recomputes from transformer data at all 6 hierarchy levels (Section → Sub-Division → Division → Circle → Zone → MESCOM)	Yes (full rebuild)	2–5 min

---

Dependency Graph

1. seed (API)
   │
   ▼
2. backfillAuditStatus ──► 3. fixInvalidAuditedTCs ──► 4. fixZeroTcConsumptionLoss
                                                            │
5. backfillBillingFlags (needs production DB connection)     │
   │                                                        │
   ▼                                                        │
6. backfillAvgConsumption (needs consumption from step 5)    │
   │                                                        │
   ▼                                                        ▼
7. backfillHealthScore (needs finalized audit fields) ◄─────┘
   │
   ▼
8. computeAuditSummary (ALWAYS LAST — depends on final TC states)

How to Run

For each script, update config.json and run:

# Example: running fixZeroTcConsumptionLoss
# 1. Edit script/config.json:
#    { "scriptRunArray": ["transformer_fixZeroTcConsumptionLoss"], ... }
#
# 2. Run:
yarn script-dev

For backfillBillingFlags (step 5), pass the production connection string in config.json input:

# 1. Edit script/config.json:
#    {
#      "scriptRunArray": ["installation_backfillBillingFlags"],
#      "input": {
#        "productionConnectionString": "mongodb+srv://user:pass@prod-host.mongodb.net/"
#      }
#    }
#
# 2. Run:
yarn script-dev

The seed (step 1) is different — it runs via the API, not the script runner:

# Start the API first, then:
curl -X POST http://localhost:3000/dtcea-mumbai-demo/v1/seed/ai-audit-test-data \
  -H "Content-Type: application/json" \
  -d '{"clearExisting": true}'

---

Important Notes

• computeAuditSummary DELETES all summaries before recomputing — always run last
• Steps 2→3→4 must be sequential — each fixes a subset of transformer audit data that the next depends on
• Step 5 (backfillBillingFlags) requires a production DB connection string — pass productionConnectionString in config.json input
• Step 5 is independent of steps 2–4 (only needs installations to exist) but must run before step 6
• Step 6 must run after step 5 — it uses the consumption field that step 5 backfills
• Step 7 must run after audit status is finalized (after steps 2–4) and after step 6
• All scripts are safe to re-run (idempotent)

Scale

• ~150K TCs × 7 months = ~1M transformer docs
• ~3.3M installation docs
• Total time per environment: ~25–45 minutes