Operator migration guide

Migrate from MikroTik User Manager to RADIUS-backed billing

A practical operator checklist for moving authentication and billing responsibilities in controlled batches, while protecting credentials, remaining credit and a tested route back.

Updated

Why operators outgrow User Manager

User Manager already provides RADIUS; this migration changes the system responsible for customer records, policy and commercial workflows. Define what your next system must prove: invoice reconciliation, local payment collection, installation handoffs and support ownership. Do not confuse an authentication success with a completed billing workflow.

Avoid a blanket claim that User Manager has no payments: MikroTik documents PayPal profile purchases under User Manager — Payments and Purchasing a profile. Check your actual requirements against the destination before retiring a working service.

This is a migration planning checklist, not a script to execute on a live router. Rehearse with synthetic accounts in an isolated lab, then approve a bounded subscriber cohort and maintenance window.

Start with the RouterOS version

RouterOS v7 User Manager uses a different database from v6. MikroTik's User Manager — Migrating from RouterOS v6 describes the manual migrate-legacy-db conversion using the complete old database folder. That converts to v7 User Manager, not to a third-party billing service.

Record the exact RouterOS and User Manager package versions on the source and each NAS (network access server). Keep the original v6 database intact if rehearsing a v7 conversion; do not overwrite a populated destination. Do not combine a router upgrade and billing cut-over in one change window. Version-specific commands and exports must be verified on a lab copy before use.

Inventory before exporting

Assign an owner to every row. Record counts, capture time, timezone, source version and an exception list. Use pseudonymous identifiers in the working inventory; keep the identity lookup restricted.

InventoryRecord without credentialsAcceptance evidence
UsersSource identity, customer/service mapping, enabled or disabled state, authentication method and duplicate usernamesEvery migrating identity has one approved destination service; duplicates are resolved within the intended tenant and realm
ProfilesAssigned profile, activation rule, expiry, price reference and pending entitlementsA customer keeps the agreed remaining entitlement
LimitationsRate, burst, quota, uptime, time-of-day restrictions, reset schedule and current consumptionA synthetic account passes each boundary test
Routers / NASSite, source address, service type, RADIUS ordering, transport, reachability and secret reference onlyAuthentication and accounting arrive from the expected NAS
Active sessionsAccount alias, NAS, session identifier, last update, address and countersSupport can distinguish stale records from currently connected customers
VouchersBatch identifier, sold/unsold/redeemed/expired status and remaining entitlementNo unsold or used voucher becomes a new free entitlement; exclude redeemable codes from the worksheet
Commercial recordsOutstanding balance, paid-through date, pending collection and reconciliation ownerFinance approves opening balances independently of access policy

The inventory is a migration recommendation, not a promise that every field has an automatic export or matching destination field. Resolve unsupported cases before moving that cohort.

Export safely: separate recovery from mapping

A text RouterOS configuration export does not contain the User Manager database. Use a separate database backup and verify a restore in isolation. See Configuration Management — Configuration Export and Import and User Manager — Database for the version-appropriate recovery procedure; v7 documents database save/load separately.

Keep two distinct deliverables:

  1. Restricted recovery archive: router configuration, User Manager database and version record. Encrypt it, limit access, record an integrity digest and restore-test it. Treat it as credential-bearing even when its filename looks harmless.
  2. Sanitized mapping worksheet: account aliases, plan references, policy values, entitlement state and exception decisions. Review the actual fields before sharing it. Do not attach raw exports to tickets or put them in Git, email or shared spreadsheets.

Passwords, NAS shared secrets, one-time-password seeds and redeemable voucher codes do not belong in the worksheet. Prefer a coordinated password reset. If continuity requires moving credentials, an authorized administrator should transfer them directly into the destination RADIUS credential store through an approved encrypted path, with access logging and temporary-file cleanup. Confirm the destination supports the actual authentication method and stored credential representation; do not assume an arbitrary password hash is portable. Test the reset or transfer with a synthetic subscriber and include CPE updates and support contact in the plan.

Set a retention deadline for temporary exports after acceptance. Preserve restricted recovery material for the agreed rollback period; redact logs and any evidence shared with reviewers.

Map profiles and limitations to plans

In v7, profiles can start on assignment or first authentication; profile limitations link schedules and limits; accounting and interim updates support usage-limit enforcement. See User Manager — Profiles, Profile Limitations, Limitations and Sessions. Validate equivalent behavior on the actual v6 source instead of assuming identical field names or semantics.

Source meaningDestination decisionLab check
Profile and user entitlementMap to a commercial plan plus service-policy revisionExisting paid time survives; assigning a plan does not restart the term
First-use or fixed expiryPreserve activation state, expiry and timezoneAn unused voucher and an already active account expire as agreed
Combined quota, uptime and scheduled limitsPreserve units, reset boundaries and consumed allowanceA nearly exhausted account does not regain a full allowance
Speed and burstsTranslate rate direction and every burst field explicitlyMeasure both upload and download, including after reconnect
Session sharing and identity restrictionsDefine the destination's concurrency and identity policyAllowed and denied second logins behave as agreed
Disabled, unpaid or held accountsCarry the restriction forward separately from plan assignmentA valid password cannot silently restore restricted access

For Mikrotik-Rate-Limit, receive is from the router's perspective: subscriber upload; transmit is subscriber download. RADIUS reply attributes can override profile defaults, while omitted attributes can leave defaults in effect. Verify the effective policy, not just a plan label. See RADIUS — Supported RADIUS Attributes: Access-Accepts packet.

Have network operations and finance approve the same mapping revision. Keep rejected or ambiguous records out of the cut-over batch.

Run both systems in parallel with clear ownership

Use separate lab or pilot NAS/cohort boundaries and one authority for each subscriber's authentication, quota and billing state. Parallel operation means comparing evidence across bounded cohorts; it does not mean two systems can both renew or charge the same account. Freeze or explicitly journal changes to the migrating cohort while comparing results.

RouterOS checks matching local access records before RADIUS, and RADIUS server ordering matters. Audit local PPP and Hotspot entries so a local account cannot mask a failed migration test. See RADIUS — Summary and RADIUS Client.

Do not treat two configured RADIUS servers as a username-based migration router. Test rejection, timeout and loss of connectivity separately. Define who receives accounting during the pilot and how the destination reconciles duplicate or late session updates. Compare authentication decisions, effective rate, address assignment, expiry and actual connectivity before expanding the cohort.

Cut-over night checklist

  1. Authorize: name the change owner, network operator, finance reviewer and support contact; publish the cohort, window, stop conditions and rollback decision time.
  2. Freeze and capture: pause conflicting renewals and policy changes for the cohort; journal unavoidable payments; take restricted recovery backups and a final sanitized count/entitlement delta.
  3. Prove the route back: confirm management access independent of subscriber authentication and the exact prior NAS settings; keep the old service available for the agreed period.
  4. Reconcile: compare destination counts, duplicates, disabled state, paid-through dates, remaining quota and voucher status against the approved mapping revision. Stop on unresolved differences.
  5. Switch the agreed boundary: have the authorized operator apply the reviewed NAS/service change. Do not bulk-disconnect the entire network. Plan whether existing sessions drain or undergo a coordinated reconnect; changing authentication targets alone is not proof that existing sessions have the new policy.
  6. Test allow and deny: exercise a valid account, wrong password, disabled account, expired entitlement and concurrency limit. Verify subscriber connectivity, assigned address and upload/download policy from the client side.
  7. Test accounting: observe session start, interim and stop evidence at the intended authority, with NAS and session identifiers. Confirm usage and remaining entitlement stay consistent after reconnect.
  8. Test commercial state separately: reconcile the journaled renewals and balances without duplicate charges or entitlement extensions. Do not infer billing correctness from an Access-Accept response.
  9. Decide and hand over: compare results with the written stop conditions. Record acceptance or rollback, the timestamp, unresolved exceptions and the next-shift owner. Expand only after the observation period agreed for this cohort.

Set measurable thresholds from your network baseline before the window: unacceptable authentication failures, missing accounting, incorrect rates or balances, and loss of management access. This guide supplies no universal outage allowance or success-rate promise.

Rollback: recover state as well as routing

If a stop condition is met, stop the rollout and freeze new writes for the affected cohort. Preserve redacted failure evidence. Restore the approved former authentication and accounting destinations for that cohort, then test a valid account, a denied account and session accounting on the old path.

Reconcile payments, voucher redemptions, expiry and consumed quota recorded since the freeze before allowing either system to resume writes. A stale database restore can erase legitimate payments or recreate spent entitlements. Replay approved deltas once, with a reconciliation owner; retain both systems' original evidence until finance and network operations sign off.

Coordinate any required subscriber credential or CPE changes with the route back. Keep a restricted record of which accounts were reset or transferred so support does not strand subscribers on the old credential set. Test existing-session behavior after rollback rather than assuming it follows new authentication automatically.

Close the window with the restored cohort, exception list, customer communications and a new rehearsal requirement. Retire the old service only after a separately approved retention and acceptance period.

ISPAgents: guided pilot scope and planned import

Planned: the ISPAgents User Manager export parser/import is not shipped; this guide does not offer a working User Manager importer.

Guided pilot: ISPAgents existing-FreeRADIUS integration provides a signed Status-Server probe and opt-in SQL accounting import into tenant session records. CoA/Disconnect drivers exist with fake-transport test coverage; real NAS mutation and end-to-end policy enforcement remain unqualified. SQL/API synchronization and policy write-back must not be treated as a complete integration.

Guided pilot: ISPAgents Managed RADIUS requires tenant-specific onboarding, live smoke verification and launch signoff; it is not a general-availability migration destination.

Guided pilot: ISPAgents customer-billing and service-policy foundations require scoped validation; a mock-safe policy workflow is not evidence of live router or RADIUS mutation.

Read the existing FreeRADIUS integration scope and FreeRADIUS versus managed RADIUS guided pilot options. Discuss a guided pilot with the sanitized inventory and unresolved mapping questions, never a password export.