liveBook · Manning

Andon cord

As explained in Build High Value Software Systems

A cord that a worker pulls to signal an abnormality and prompt immediate attention or correction.

An Andon cord makes a problem visible early so it can be corrected before it spreads. In software, divergence between code and its specification is an Andon moment: pause long enough to resolve the mismatch using the specification’s decision boundaries. Escalate only when the intent itself is ambiguous. section 4.1.4

In production, pulling the cord does not mean stopping the whole system. Each failure calls for a deliberate action, such as rolling back a bad deploy, pausing a resource-heavy job, or disabling a problematic feature flag. Define these actions in advance: the wrong intervention can make the problem worse. section 12.3.2

Listing 4.4 Spec file
Accept: GET /api/orders/export (authenticated user)
Return: CSV file with columns: order_id, date, total, status
Filter: Only orders belonging to the requesting user
Format: Standard CSV, UTF-8 encoding
Never: Include other users' data, build the CSV in the browser, load all orders into memory, allow unauthenticated access
Prefer: Stream CSV rows from the server using bounded database reads

A concise export specification makes expected behavior explicit, giving the team a basis for spotting divergence. section 4.1.4

Table 12.3 What "pull the cord" means in practice
Failure mode
Andon action (the cord)
What it stops
What keeps running
Implication if wrong
Bad deploy causing errors
Roll back to previous version
The defective code path
All traffic on the known good version
Rollback introduces its own change; must be tested
Dependency failure (API, DB)
Trip circuit breaker
Calls to the failing dependency
Degraded service with fallback or cached data
Users see partial functionality, not full service
Traffic spike overwhelming capacity
Shed load via rate limiting
Excess requests beyond capacity
Priority traffic within safe limits
Some legitimate users get rejected
Feature flag causing latency
Disable the flag
The new code path
All traffic on the old code path
Assumes old path still works; must not have been removed
Batch job exhausting shared resource
Pause the batch job
Resource consumption by the job
All user-facing traffic unaffected
Batch data arrives late; downstream reports may run against incomplete data
Data corruption from migration
Halt pipeline, freeze writes
Further data modification
Read traffic continues; writes queue or fail
Extended read-only mode affects transactional workflows

The table maps failure modes to specific actions, what each action stops, and the risk if it is wrong. section 12.3.2

How Build High Value Software Systems covers it

How deep it goes
Defines the signal and details correction boundaries, alert causes, and preplanned production actions without treating production like a stoppable factory line.
Examples
Uses code-specification drift and maps production failures to rollback, circuit-breaker, rate-limit, feature-flag, or batch-job actions.
What you can do after
Apply it by detecting specification drift early or choosing a preplanned, cause-specific production action instead of stopping the whole system.
How it connects
Connects specifications and quality gates with alerts, feature flags, rollbacks, circuit breakers, and rate limiting.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build High Value Software Systems. The section links point to the parts of the book it draws on.