Skip to content

Daily Checks

Applies to: All subscriptions

Purpose

Catch, in ten to fifteen minutes each working day, the problems that accumulate silently — before they become a compliance failure or a user complaint.

Audience

The customer's operations owner, and their deputy.

Prerequisites

Steps

Work in this order. It runs from the most time-critical to the least.

1. Yesterday's alerts
Verify Every alert raised since the last check has been acted on, not merely acknowledged
Expect No open alerts without an owner and a next action
Warning signs The same alert recurring daily; alerts acknowledged with no change
Act Assign an owner. If an alert recurs without action, either fix the cause or delete the alert — a permanently ignored alert devalues the whole channel

2. Documents ageing towards the reporting window

The most important check in this routine, because it is the only one whose deadline cannot be recovered.
Verify The oldest document awaiting registration, per registration
Expect Comfortably inside the reporting window
Warning signs Anything past half the window; a count that grew since yesterday
Act Register them today. If blocked by a provider problem or a validation fault, escalate now — not tomorrow. Once the window closes there is no remedy

3. The correction queue
Verify How many documents are quarantined, and how old the oldest is
Expect Yesterday's cleared; today's being worked
Warning signs Growing day on day; anything more than three days old; many documents failing the same way
Act Clear the queue. If many share one cause, fix the source — the mapping template or the master data. Correcting them one at a time will not keep pace

4. Failed generations
Verify Documents that failed registration since the last check
Expect Few, and each with a specific cause
Warning signs A rising count; the same code repeatedly; failures with no code at all
Act Look up each code and act by its classification. Failures with no code point at the provider — run the connection test. See IRN Failures

5. e-Way Bills approaching expiry
Verify Bills expiring within the alert horizon
Expect Each has an owner and a plan
Warning signs Bills expiring with goods still in transit; a recurring pattern of near-misses
Act Extend where possible and inside the window. A recurring pattern means the horizon is too short — set it from actual lead times

6. Upload and background jobs
Verify Jobs submitted yesterday completed
Expect All complete
Warning signs Jobs still queued; a queue depth that only rises
Act Check the worker and the queue. This does not show as an error in the interface — the application looks entirely healthy while nothing is processed

7. Integration feeds
Verify Each ERP sent roughly its normal volume
Expect Within the normal range for the day of the week
Warning signs Zero from any ERP during business hours; a sharp fall
Act Check the ERP side first — Complifly cannot distinguish a stopped integration from a quiet day. See Ingest Failures

8. Write-back backlog
Verify The age of the oldest pending event
Expect Below the consumer's normal interval
Warning signs Age rising; dead-lettered events present
Act Check the consumer, then the endpoint. Diagnose on age, not count — a steady count hides a stuck queue

9. System health
Verify Application responding, disk space, no unexpected restarts
Expect Healthy
Warning signs Restarts nobody scheduled; disk below the warning threshold
Act Investigate restarts — they are rarely benign. Free or extend disk before it becomes urgent

The five-minute version

When time is short, do these three. They are the ones with deadlines:

  1. Documents ageing towards the reporting window
  2. e-Way Bills approaching expiry
  3. Integration feeds at zero

Validation

Check Pass condition
The routine was performed Logged, with the date and who did it
Findings have owners Each with a name and a date
Nothing is ageing Oldest quarantined document, oldest unregistered document, oldest pending event all within limits
Yesterday's findings closed Or carried forward deliberately
The deputy can run it Demonstrated at least once

Troubleshooting

Symptom Cause Action
The routine takes over 30 minutes Backlogs have accumulated Clear them once, then keep them clear. The daily routine is cheap only when nothing is stuck
Quarantine grows despite daily clearing The cause is systemic Fix the mapping or master data
Documents found past the reporting window The ageing check was skipped It cannot be recovered. Make it the first check every day
Jobs queued and nobody noticed Not checked, and no error surfaces Add job completion to the routine and to monitoring
An integration stopped days ago Feed volume not checked Check every ERP, every day
Alerts acknowledged, nothing changes No follow-up owner Assign owners at the point of acknowledgement
The routine lapses during leave No deputy Name and train one