Project 8 — Advanced

REFramework Production Bot (Orchestrator ke saath)

Ab tak ke projects development stage ke the. Enterprise me bot ko 24x7 production me chalna hai — queues, retries, alerts, monitoring ke saath. Is project me REFramework (Robotic Enterprise Framework) samjhenge aur ek production-grade bot banayenge. Ye RPA developer ki highest-value skill hai.

⏱ 35 min• Advanced• UiPath Studio (Community)

REFramework kya hai?

WHAT

UiPath ka official blueprint jo production bot ko ek state machine me organize karta hai — Initialization → Get Transaction Data → Process Transaction → Reconciliation. Queues, assets, retries, exception handling sab framework me built-in. Enterprise bots isi structure me bante hain.

WHEN

Jab bot production me jaye — 24x7 runs, multiple robots, business users ko alerts chahiye. Demo/POC me simple sequences chalte hain, real deployment me REFramework mandatory hai (client requirement hoti hai).

WHERE

Har enterprise RPA project — banking, insurance, healthcare, telecom. Job descriptions me "REFramework experience" likha hota hai. Isko samajhne se senior RPA developer bante ho.

4 states ka architecture

REFramework State Machine:

INITIALIZATION
 +-- InitAllSettings (Config.xlsx read -> Settings dict)
 +-- Init queues, assets, directories
 +-- Get Transaction Data
      |
      v
GET TRANSACTION DATA
 +-- Get Transaction Item  (Orchestrator queue se next item)
 +-- TransactionItem = Nothing? -> go RECONCILIATION (queue khaali)
 +-- TransactionItem mila? -> PROCESS TRANSACTION
      |
      v
PROCESS TRANSACTION
 +-- Try Catch
 |    +-- Process.xaml invoke  (asli kaam - item process)
 |    +-- Set Transaction Status: Successful
 |    +-- (retries available? agar business exception aaya)
 +-- Catch (Business Exception)
 |    +-- Set Transaction Status: Failed / BusinessException
 |    +-- (data galat hai - retry ka fayda nahi)
 +-- Catch (System Exception)
 |    +-- retry count check
 |    +-- Set Transaction Status: Failed  (auto-retry by Orchestrator)
 +-- (loop back to GET TRANSACTION DATA)
      |
      v
RECONCILIATION
 +-- InitAllSettings again, cleanup
 +-- End Process  (bot band)

Config.xlsx — bot ka brain

Config.xlsx sheets:

Settings sheet:
Key                          Value
RETRIES_NUMBER               3
SCREENSHOTS_PATH             C:\Screenshots\
EXCEPTION_SCREENSHOTS_FOLDER C:\Screenshots\Exceptions\
QUEUE_FOLDER_NAME            Finance
MAX_RETRIES                  3

Assets sheet (Orchestrator asset mapping):
AssetName            AssetType
Credential_DB        Credential
CRM_ClientId         Text
SMTP_Password        Credential

Queues sheet:
QueueName              REF_Invoice_Processing

# InitAllSettings ye sab load karta hai - code me ek bhi hardcode nahi

Process.xaml me asli kaam

Process.xaml (In/Out Arguments: in_TransactionItem, out_ok)

 +-- Try Catch
 |    +-- Assign: strData = in_TransactionItem.SpecificContent("InvoiceJson")
 |    +-- Sequence: Parse + Validate
 |    |    +-- Deserialize JSON -> dtInvoice
 |    |    +-- If  dtInvoice.Total <= 0  -> Throw  "Invalid total" (Business)
 |    +-- Sequence: Execute
 |    |    +-- HTTP Request / Excel write / DB insert (asli processing)
 |    +-- Catch (Business Exception)
 |    |    -> Throw (data issue, framework Failed/BusinessException me mark karega)
 |    +-- Catch (System Exception)
 |         -> Take Screenshot  SCREENSHOTS_PATH
 |         -> Rethrow  (framework retry count check karega)

Business vs System exception — interview ka favourite

Business Exception = DATA galat hai
  - Invoice total 0 hai, email format invalid, mandatory field missing
  - Retry se fayda nahi -> Failed (BusinessException), manual queue me jaata hai
  - Set Transaction Status: Failed / BusinessException

System Exception = ENVIRONMENT ka issue
  - Network down, API 500, DB locked, file missing
  - Retry se theek ho sakta hai -> Orchestrator auto-retries (RETRIES_NUMBER)
  - Screenshot log karo, Rethrow

# Rule of thumb: "Kya dobara same item try karne se theek hoga?"
# Haan -> System. Nahi -> Business.

Orchestrator operations

Queues: Dispatcher bot items daalta hai, Performer bot leta hai — scale independent. 1 dispatcher → 10 performers.
Assets: credentials, paths, config values — bot redeploy bina change.
Triggers: time-based + event-based (file trigger). Trigger bot = queue pe item aaye to automatically start.
Monitoring: Orchestrator dashboard — running jobs, failed transactions, queue depth. Alerts configure karo (job fail → email/Slack).
Folders: environment separation — DEV/QAS/PRD folders. Prod bot ko QAS queue kabhi na dikhe.

Interview question: "REFramework me kyun use karte ho?" — Production stability ke liye: built-in retries, exception classification, queue processing, config management, logging standard. Client RPA CoE me har bot isi framework pe hota hai — custom framework review me reject ho jata hai. Aur "transaction retry kaise hota hai?" — System exception pe RETRIES_NUMBER count hota hai, max tak auto-retry, phir dead-letter.

Exercise

Question: Orchestrator queue se next transaction item laane wali activity? (3 words)

Question: Data galat hone par (invoice total 0) kaunsa exception maara jaata hai? (2 words)

Common mistakes

Har exception ko same treat karna: Business aur System exception ka farq samjho — warna retries bekaar ya data loss.
Hardcoded config: RETRIES_NUMBER code me likha = production tuning ke liye redeploy. Config.xlsx.
No screenshots: System exception pe screenshot na lena = baad me "kya hua" pata nahi. Screenshot + log mandatory.
Transaction me cleanup skip: Process ke end me temp files/cookies clear karo — warna next transaction pe pollution.
DEV config on PRD: Folder separation (DEV/QAS/PRD) bina production run = disaster. Orchestrator folders use karo.

Lesson complete?

Course complete! 8 real automation projects tumhare portfolio me. Ab inhe apne resume me daalo aur RPA interviews start karo.