Audit production-configuration-and-approval-methodology.md thresholds; point to source-of-truth and consolidate defaults

[BigLep]

Summary

Audit the accuracy of the threshold/parameter values in docs/checks/production-configuration-and-approval-methodology.md, make the doc point at the source(s) of truth rather than restating values that can drift, and reduce the number of places a given default is defined.

Motivation

The doc currently restates configuration values inline, and at least one of them does not match (or is not actually enforced by) the code. Example found while investigating the Data Storage Success Rate section:

• The doc lists Max dataStorageCheckMs = 180s.
• There is no 180 anywhere in apps/backend/src (outside unrelated Prometheus histogram bucket boundaries), no config default of 180, and no env/infra override setting it to 180.
• What actually causes the data storage ("deal") check to give up is the AbortController + setTimeout in jobs.service.ts handleDealJob, driven by dealJobTimeoutSeconds, which defaults to 360s (Math.max(120000, timeoutSeconds * 1000) floor of 120s), and is set to 360 in prod env.

So the doc implies a 180s hard cutoff that does not exist in code. Whether 180s is a stale value, an aspirational SLO, or a genuinely intended cutoff that was never implemented needs to be resolved as part of the audit. This is exactly the kind of drift that happens when values are copied into prose instead of referenced.

Scope of work

1. Audit doc values and link to the source of truth

Go through every threshold/parameter value in docs/checks/production-configuration-and-approval-methodology.md, confirm each against the code/config, and where practical replace the inline value with a link to (or clearly reference) the source of truth. Candidate sources of truth:

• Code defaults: apps/backend/src/config/constants.ts
• Env-loading defaults: apps/backend/src/config/loader.ts
• Prod overrides: prod/apps/dealbot/dealbot-env.env (FilOzone/infra)

Resolve the dataStorageCheckMs = 180s discrepancy specifically (correct the value, remove it, or file follow-up to actually enforce it, whichever is intended).

2. Define each default in exactly one place

Today defaults live in two places:

• constants.ts networkDefaults (per-network defaults, e.g. dealJobTimeoutSeconds: 360).
• loader.ts inline getNumberEnv(env, "X", <default>) calls for non-per-network config (e.g. connectTimeoutMs: 10000, httpRequestTimeoutMs: 240000, ipniVerificationTimeoutMs: 60000, sampledBlockSampleCount: 5, etc.).

Consolidate so each default value is declared once (ideally all in constants.ts, with loader.ts referencing those constants) so there is a single, greppable source of truth.

3. Add "keep the doc in sync" reminders next to the values

Add comments in both:

• apps/backend/src/config/constants.ts
• prod/apps/dealbot/dealbot-env.env (FilOzone/infra)

noting that changes to these values likely also require an update to docs/checks/production-configuration-and-approval-methodology.md.

Notes

• Item 3's env-file change lives in the separate FilOzone/infra repo, so it will need a companion PR there (or a linked follow-up issue).

[BigLep]

Note: I was triggered to create this issue when looking at https://github.com/FilOzone/dealbot/blob/main/docs/checks/production-configuration-and-approval-methodology.md with fresh eyes, knowing Curio maintainers are coming up to speed on the dealbot's "rules of the game" as part of working on FilOzone/filecoin-services#546