d365-flow-reliability
/d365-flow-reliability
Projektuje niezawodne przepływy Power Automate dla Dataverse i Dynamics 365: filtrowanie wyzwalaczy, obsługę błędów, ponowienia i idempotentność.
- Autor
- antonioddiniz ↗
- Repozytorium
- d365-skills ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 30.10.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 2
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 8
Brak sygnałów. Repozytorium zmienione 30.10.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Projektuje niezawodne przepływy Power Automate dla Dataverse i Dynamics 365: filtrowanie wyzwalaczy, obsługę błędów, ponowienia i idempotentność.
Wyzwalacze z opisu
Oryginalny opis (EN)
Make Power Automate cloud flows that touch Dataverse / Dynamics 365 reliable and cheap to run - trigger filtering (select columns, filter rows, trigger conditions, scope), try/catch/finally with Scope and Configure run after, retry policies, idempotency against duplicate triggers, Apply to each concurrency, List rows paging and FetchXML, throughput limits, connection references and run-only users. Use this skill whenever the user designs or reviews a flow on "When a row is added, modified or deleted", a flow that runs too often or loops, fails silently, creates duplicates, hits throttling or 5
Jak działa
- Trigger: run only when it matters
- Failure: try, catch, finally
- Retries and idempotency
- Volume
- Portability
- Review checklist
- Output
- References in this skill
- Sources (Microsoft Learn)
Podgląd SKILL.md
---
name: d365-flow-reliability
description: Make Power Automate cloud flows that touch Dataverse / Dynamics 365 reliable and cheap to run - trigger filtering (select columns, filter rows, trigger conditions, scope), try/catch/finally with Scope and Configure run after, retry policies, idempotency against duplicate triggers, Apply to each concurrency, List rows paging and FetchXML, throughput limits, connection references and run-only users. Use this skill whenever the user designs or reviews a flow on "When a row is added, modified or deleted", a flow that runs too often or loops, fails silently, creates duplicates, hits throttling or 5,000-row limits, needs error notifications, or will be moved to another environment - even if they only paste a flow definition JSON or a screenshot description.
---
# Reliable Dataverse flows
A flow that triggers on every save of a busy table is a plug-in without a debugger. Filter first,
then handle failure, then make re-runs harmless.
## Trigger: run only when it matters
1. **Change type** only for the events you handle (Added, Modified, Deleted).
2. **Select columns** (Modified only) to the columns that matter. The flow fires when a listed
column is *in the update request*, even with the same value. Never list the primary key or
columns that every update sends. Lookup columns are not supported in Select columns.
3. **Filter rows** with an OData expression (no `$filter=` prefix) so the flow runs only when
the row qualifies after the save: `statecode eq 0 and revenue gt 100000`.
4. **Trigger conditions** for logic OData cannot express. They stop the run before it costs anything.
5. **Scope** (User, BU, Parent:Child, Organization) is about whose rows trigger the flow.
Integrations usually need Organization.
6. **Loop guard:** a flow that updates the same row it triggers on must not list that column in
Select columns, or must compare before writing.
## Failure: try, catch, finally
- Put the work in a **Scope "Try"**. Add **Scope "Catch"** with Configure run after =
has failed / has timed out. Add **Scope "Finally"** after both if cleanup is needed.
- In Catch, read the error with `result('Try')` filtered to failed actions, log it with the run
URL built from `workflow()`, then **terminate as Failed** so monitoring sees the failure.
- Do not catch and continue silently. A green run that did nothing is the worst outcome.
## Retries and idempotency
- Connector actions retry by default (exponential, up to 4 retries for most). For calls that hit a
slow dependency, set an explicit exponential policy with longer intervals.
- **Retries and duplicate triggers happen.** Make writes idempotent: upsert by alternate key,
check-before-create, or store a processed marker. Never "Add a new row" blindly in a retried path.
- 429 from Dataverse is handled by the connector retry. Do not add a "Delay 1 minute and try again"
loop on top of it.
## Volume
- **List rows** returns up to 5,000 rows by default. Turn on pagination with a threshold or loop on
the skip token for more, and always set Select columns and Filter rows.
- **Apply to each** runs sequentially by default. Concurrency (1 to 50) helps only at the top level:
nested loops run sequentially. Never use concurrency with variables you append to.
- Prefer one FetchXML query with aggregates or a Dataverse bulk/Custom API call over thousands
of loop iterations.
- Throughput limits depend on the flow's performance profile. A flow that stays above them for
14 days is turned off.
## Portability
- Flows live in solutions, use **connection references**, and read URLs and ids from
**environment variables**. No hard-coded GUIDs, environment URLs or personal connections.
- The owner is a service account or service principal, not the person who built it.
## Review checklist
- [ ] Trigger has Change type, Select columns and Filter rows set, or a stated reason why not?
- [ ] No self-trigger loop?
- [ ] Try/Catch/Finally scopes, Catch terminates as Failed and logs the run URL?
- [ ] Writes are idempotent under retry and duplicate trigger?
- [ ] List rows has Select columns, Filter rows, and paging if it can exceed 5,000?
- [ ] Concurrency used only where order and shared variables do not matter?
- [ ] Connection references, environment variables, service-account owner?
## Output
Return the issues by severity with the exact setting to change (trigger property, action setting,
expression), and the corrected trigger and Catch scope.
## References in this skill
- `references/expressions.md` - trigger conditions, error extraction, run URL, idempotency patterns.
## Sources (Microsoft Learn)
- Dataverse trigger (filter columns, filter rows, scope): https://learn.microsoft.com/power-automate/dataverse/create-update-delete-trigger
- Optimize triggers (trigger conditions): https://learn.microsoft.com/power-automate/guidance/coding-guidelines/optimize-power-automate-triggers
- Robust error handling (scopes, run after, retry): https://learn.microsoft.com/power-automate/guidance/coding-guidelines/error-handling
- Work only with relevant data: https://learn.microsoft.com/power-automate/guidance/coding-guidelines/work-with-relevant-data
- Parallel execution and concurrency: https://learn.microsoft.com/power-automate/guidance/coding-guidelines/implement-parallel-execution
- List rows (paging, FetchXML): https://learn.microsoft.com/power-automate/dataverse/list-rows
- Limits and configuration: https://learn.microsoft.com/power-automate/limits-and-config
- Default retry policy behavior: https://learn.microsoft.com/azure/logic-apps/error-exception-handling#retry-policies
Pliki w paczce (2)
- SKILL.md 6 KB
- references/expressions.md 1 KB
Wersje
W bazie nie ma innych skilli o nazwie d365-flow-reliability.
Migawka bazy z 13.11.2026 · wersja paczki ada08a6cad30 (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100power-automate-solution-flow-standards
Ustala standardy budowy przepływów Power Automate w rozwiązaniu Dataverse: nazwy, referencje połączeń, obsługę błędów i wdrożenia. - 100dataverse-flow-api
Tworzy, podłącza połączenia, aktywuje i poprawia przepływy Power Automate przez Dataverse Web API, np. dla formularzy Microsoft Forms zapisujących wiersze. - 100dynamics365-expert
Wspiera pracę z platformą Dynamics 365: aplikacje model-driven i canvas, przepływy Power Automate, Dataverse, wtyczki i niestandardowe workflow. - 90dataverse-uat-plan
Buduje plan testów akceptacyjnych dla zmian w Dynamics 365, Dataverse i Power Automate oraz rzetelnie prowadzi status każdego testu. - 100power-platform
Buduje, wdraża i weryfikuje rozwiązania Microsoft Power Platform jak oprogramowanie: aplikacje canvas jako .pa.yaml w git, Dataverse, przepływy Power Automate, pac pack oraz strony Power Pages. - 90flow-design-patterns
Ustala wzorce budowy niezawodnych przepływów Power Automate z krokami AI: wyzwalacze, referencje połączeń, zmienne, walidację wyników i obsługę błędów.