使用公司数据自扩展常见问题
备注
2020 年 11 月生效：
- Common Data Service 已更名为 Microsoft Dataverse。 有关详细信息，请参阅 Power Automate 博客。
- Microsoft Dataverse 中的一些术语已更新。 例如，实体 现在更新为 表，字段 现在更新为 列。 有关详细信息，请参阅术语更新。
本主题很快将更新以反映最新的术语。
为什么需要自扩展？
您可能已有一个包含业务数据的 Dataverse 或其他 Dynamics 365 应用实例，并且想要对其启用双写入连接。 在这种情况下，您需要在启用双写入连接之前使用公司信息自扩展 Dataverse 或其他 Dynamics 365 应用数据。
我什么时候应该使用自扩展？
应在启用双写入表映射之前（步骤 #5）使用自扩展。
要在 Finance and Operations 应用的实例与 Dataverse 或其他 Dynamics 365 应用之间设置双写入连接，请以管理员身份登录到 Finance and Operations 应用。
转到 数据管理 模块，然后单击 双写入 按钮。 这将启动 数据集成器。
为一个或多个公司创建双写入连接。
启用 Cdm_companies 表映射。 这会使公司从 Finance and Operations 应用同步到 Dataverse。
在 Dataverse 或其他 Dynamics 365 应用实例上运行示例自扩展代码。
当自扩展完成并且系统准备好实时同步后，启用表映射。
启用表映射会触发已启用表映射的初始数据同步。 与在双写入连接上选择的公司相对应的数据将在 Finance and Operations 应用与 Dataverse 之间同步。
如何使用代码示例？
示例代码是一个 C# 应用程序，您可以在 Visual Studio 中加载它。 它获取 Dataverse 上的 NuGet 包依赖项，您可以通过标准 Visual Studio 工具进行刷新。
解压缩并在 Visual Studio 中打开解决方案并还原 NuGet 包后，在代码中搜索 TODO。 TODO 指出了针对您要如何自扩展公司信息需要做的每个决策，以及用于规范实现的示例代码。
示例代码仅显示您可以按公司对实体行进行分类的多种方式中的一种。 通过更改 TODO 部分的逻辑，您可以创建自定义分类。
我会看到什么结果？
默认情况下，示例应用程序使您可以提供业务单位到公司代码映射的字典。 您使用 OwningBusinessUnit 字段进行自扩展的任何实体都会自动设置为使用指定的公司。 没有 OwningBusinessUnit 字段的任何实体（例如产品）都将根据映射使用空业务单位值设置公司。
控制台应用程序需要一个参数，即 –simulate 或 –apply。 如果您使用 –simulate 命令行参数，则不会更新数据。 在与工具相同的目录中仅生成 simulation_.csv 文件，每个更新的实体一个。 您可以在确保代码按预期更新公司值的同时迭代地查看这些文件。
完成模拟更新后，使用 –apply 参数。 这将更新当前具有不正确公司值的所有行，按一次更新 1000 个行的批次进行（默认）。 所提供的代码是幂等的，这意味着您可以重新运行它，并且仅会更新错误分配的公司。 当使用 –apply 运行时，代码将输出所做更改的 CSV 文件，这些文件名为 applied_.csv。
using Microsoft.Crm.Sdk.Messages;
using Microsoft.Xrm.Sdk;
using Microsoft.Xrm.Sdk.Messages;
using Microsoft.Xrm.Sdk.Query;
using Microsoft.Xrm.Tooling.Connector;
using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.IO;
namespace BootstrapCompany
{
/// <summary>
/// Application to bootstrap the company field on existing rows in CDS in preparation for integration to Finance and Operations.
/// </summary>
/// <remarks>
/// This application assumes that the target companies already exist in the CDS environment in the cdm_Company table and are
/// identified by their company code. It also assumes that the current owning business unit of each row should be used
/// to categorize by company. This logic can easily be updated to utilize alternate sources of categorization including
/// custom tables, teams, custom fields on tables, or any other data. This code is provided only as a sample.
///
/// To utilize this code, update each of the locations currently denoted with a TODO statement.
///
/// This code is provided AS IS with no warranties or guarantees, and confers no rights.
/// </remarks>
public class Program
{
/// <summary>
/// The number of rows to query and update in CDS in a single operation.
/// </summary>
/// <remarks>
/// The larger this number, the fewer calls will need to be made, so the faster the updates
/// will complete. However, larger batch sizes are more likely to cause contention. Additionally,
/// when SQL exceeds some threshold of locks (generally around 5,000), it will escalate to
/// an entire table lock, which blocks all other activity in the live system on this table. As
/// such, a batch size of around 1,000 is relatively fast, while also relatively safe in terms
/// of contention and transaction time.
/// </remarks>
const int requestBatchSize = 1000;
/// <summary>
/// The number of faults that may be seen in CDS before the operation is aborted and an exception is thrown.
/// </summary>
/// <remarks>
/// An occassional error due to contention when updating large tables in production is expected, so by default
/// errors are logged and skipped. However, if a large number of errors are seen, ignoring those errors
/// in subsequent batches gets expensive, and is usually indicative of a larger issue that should be addressed
/// before continuing. Faulted requests are *not* retried, but would be picked up in a subsequent run of this script.
/// </remarks>
const int maxFaultThreshold = 100;
/// <summary>
/// The maximum number of rows per business unit to export when simulating.
/// </summary>
/// <remarks>
/// During simulation, queries are not batched since doing so would require ordering and so be slightly
/// different from the actual execution logic. To keep this the same between both paths, simulates are
/// not batched and so a separate maximum number of rows per business unit can be specified.
/// </remarks>
const int maxSimulateRecordsPerBusinessUnit = 10000;
/// <summary>
/// Whether or not operations should continue if any errors are encountered.
/// </summary>
/// <remarks>
/// This is different than setting maxFaultThreshold = 0, since the first batch of updates will be processed
/// together. If continueOnError is true and maxFaultThreshold is 0, it is possible that multiple errors may
/// be encountered and at the same time some rows successfully updated. In a healthy system when updating
/// a higher number of rows, an occasional spurious error is expected, so it is recommended this be left as true.
/// </remarks>
const bool continueOnError = true;
#region private variables
private static Dictionary<string, EntityReference> cachedCompanyReferences = new Dictionary<string, EntityReference>();
#endregion
/// <summary>
/// The main execution loop of the program.
/// </summary>
/// <param name="args">No arguments are expected.</param>
static void Main(string[] args)
{
if (args.Length != 1 && args[0] != "-simulate" && args[0] != "-apply")
{
Console.WriteLine("Usage: BootstrapCompany -simulate");
Console.WriteLine(" BootstrapCompany -apply");
Console.WriteLine("The -simulate flag will create a file called simulation.csv in the working");
Console.WriteLine("directory, but will not change any data. The -apply flag will update live data");
Console.WriteLine("in the same way that was demonstrated in the simulation.");
return;
}
bool isSimulate = args[0].Equals("-simulate", StringComparison.OrdinalIgnoreCase);
// Delete the simulation or applied files if existing
foreach (string existingSimulate in Directory.EnumerateFiles(Directory.GetCurrentDirectory(), $"{(isSimulate ? "simulation" : "applied")}_*.csv"))
{
File.Delete(existingSimulate);
}
IOrganizationService orgService;
// TODO: Provide your connection string details for your environment
CrmServiceClient cdsConnection = new CrmServiceClient("AuthType=Office365;Username=youraliashere@yourdomainhere.com;Password=yourpasswordhere;URL=https://yourorganizationurlhere.crm.dynamics.com/;");
orgService = (IOrganizationService)cdsConnection.OrganizationWebProxyClient != null ? (IOrganizationService)cdsConnection.OrganizationWebProxyClient : (IOrganizationService)cdsConnection.OrganizationServiceProxy;
if (orgService != null)
{
// Get the current user ID to verify the connection was successful
Guid userid = ((WhoAmIResponse)orgService.Execute(new WhoAmIRequest())).UserId;
if (userid != Guid.Empty)
{
Console.WriteLine("Connection Successful!");
}
// TODO: Provide a mapping of OwningBusinessUnit name to cdm_Company company ID. You can reuse
// the same company ID for multiple business units if desired. In this example, it assumes that
// the business unit named "USMF" is related to the company "USMF". If all rows were owned
// by the same root business unit, then the first field in the dictionary should be set to the
// name of the root business unit, usually the same value as the organization (eg, "Contoso").
Dictionary<string, string> businessUnitToCompanyMapping = new Dictionary<string, string>()
{
{ "", "USMF" }, // The default mapping to use for any entity that doesn't have an owningbusinessunit field
{ "USMF", "USMF" },
{ "FRRT", "FRRT" },
};
// TODO: Provide a list of tables for which the company field should be backfilled based
// on owning business unit. The list below represents all existing tables for which a cdm_Company
// lookup field was added as part of the Finance and Operations dual write project.
BatchUpdateEntity(orgService, "account", "msdyn_company", businessUnitToCompanyMapping, true, isSimulate, "accountnumber", "name");
BatchUpdateEntity(orgService, "contact", "msdyn_company", businessUnitToCompanyMapping, true, isSimulate, "fullname");
// ... Add more here
// Note, the product entity does not have an owningbusinessunit field like most other tables, so
// assigning company by Business Unit is not applicable. In this case, whichever mapping specifies an
// empty business unit will be used to categorize tables without an owningbusinessunit field.
BatchUpdateEntity(orgService, "product", "msdyn_companyid", businessUnitToCompanyMapping, false, isSimulate, "productnumber");
}
else
{
Console.WriteLine("Connection failed...");
}
Console.WriteLine("Done");
Console.ReadLine();
}
/// <summary>
/// Updates all incorrectly assigned company relationships for the specified entity.
/// </summary>
/// <param name="orgService">The connection to CDS.</param>
/// <param name="entityName">The logical name of the entity to update.</param>
/// <param name="companyFieldName">The physical name of the field in the entity being updated which contains the cdm_Company id.</param>
/// <param name="businessUnitToCompanyMapping">A dictionary of business unit name to company code.</param>
/// <param name="hasOwningBusinessUnit">true if the entity has an owningbusinessunit field; otherwise, false.</param>
/// <param name="isSimulate">true to simulate output; otherwise, false.</param>
/// <param name="fieldsToExport">A set of fields to export into a CSV for this entity if simulating.</param>
/// <returns>true if the entity was successfully processed without any errors; otherwise, false.</returns>
private static bool BatchUpdateEntity(
IOrganizationService orgService,
string entityName,
string companyFieldName,
Dictionary<string, string> businessUnitToCompanyMapping,
bool hasOwningBusinessUnit,
bool isSimulate,
params string[] fieldsToExport)
{
List<Guid> faultedIds = new List<Guid>();
int totalRecordsProcessed = 0;
Stopwatch stopwatch = new Stopwatch();
stopwatch.Start();
string fileName = isSimulate ? "simulation" : "applied";
StreamWriter simulationWriter = new StreamWriter(Path.Combine(Directory.GetCurrentDirectory(), $"{fileName}_{entityName}.csv"), true);
simulationWriter.Write("EntityName,EntityId,");
foreach (string fieldToExport in fieldsToExport)
{
simulationWriter.Write($"{fieldToExport},");
}
simulationWriter.WriteLine("BusinessUnit,NewCompanyId");
// Process each mapped business unit individually
foreach (string businessUnitName in businessUnitToCompanyMapping.Keys)
{
Console.WriteLine("Updating any {0} rows for business unit {1} to company {2}...", entityName, businessUnitName, businessUnitToCompanyMapping[businessUnitName]);
// The empty business unit value is only applicable for tables without an owning business unit field
if (hasOwningBusinessUnit && string.IsNullOrEmpty(businessUnitName))
{
continue;
}
else if (!hasOwningBusinessUnit && !string.IsNullOrEmpty(businessUnitName))
{
continue;
}
var companyRef = GetCompanyReference(orgService, businessUnitToCompanyMapping[businessUnitName]);
// Iteratively loop in batches to keep transaction lock size small
bool moreRecordsExist = true;
while (moreRecordsExist)
{
moreRecordsExist = false;
// Find the first batch of rows for this business unit with the wrong company ID. Ordering
// is not explicity specified, but SQL will most likely process based on the index starting with
// company ID, since all new company ID fields added for Finance and Operations integration have
// also added a new index starting with company ID. Explicitly specifying order would reduce the
// query plan options for SQL and introduce unnecessary overhead.
QueryExpression query = new QueryExpression(entityName);
query.ColumnSet.AddColumns(companyFieldName);
foreach (string fieldToExport in fieldsToExport)
{
query.ColumnSet.AddColumn(fieldToExport);
}
query.Criteria.AddCondition(companyFieldName, ConditionOperator.NotEqual, companyRef.Id);
// TODO: Uncomment the line below if you only want to fill in companies that are empty
// as opposed to the line above which updates the company any time it differs from the
// desired value
// query.Criteria.AddCondition(companyFieldName, ConditionOperator.Equal, Guid.Empty);
if (isSimulate)
{
// During simulation, get as a single block of rows to avoid positioning complexities
query.TopCount = maxSimulateRecordsPerBusinessUnit;
}
else
{
// Only batch rows during actual application, otherwise retrieve all as a single operation
query.TopCount = requestBatchSize + faultedIds.Count;
}
// For tables with an owning business unit, join based on business unit name
if (hasOwningBusinessUnit)
{
// TODO: Replace this logic with different algorithms to determine the correct company
// in situations where business unit is not the best way to categorize.
LinkEntity linkEntity = query.AddLink("businessunit", "owningbusinessunit", "businessunitid", JoinOperator.Inner);
linkEntity.Columns.AddColumns("name");
linkEntity.LinkCriteria.AddCondition("name", ConditionOperator.Equal, businessUnitName);
}
var multipleRequest = new ExecuteMultipleRequest()
{
Settings = new ExecuteMultipleSettings()
{
ContinueOnError = true,
ReturnResponses = true
},
Requests = new OrganizationRequestCollection()
};
EntityCollection result = orgService.RetrieveMultiple(query);
int rowsAddedToBatch = 0;
foreach (var entity in result.Entities)
{
// Skip any previously faulted ID's. These values will be re-queried with each batch
// which is inefficient, but is more efficient than passing hundreds of ID values to
// the underlying SQL query to be skipped at the database level (assuming the
// max fault count is relatively small).
if (faultedIds.Contains(entity.Id))
{
continue;
}
entity.Attributes[companyFieldName] = companyRef;
UpdateRequest updateRequest = new UpdateRequest()
{
Target = entity
};
simulationWriter.Write($"{entityName},{entity.Id},");
foreach (string fieldToExport in fieldsToExport)
{
simulationWriter.Write($"{entity.Attributes[fieldToExport]},");
}
simulationWriter.WriteLine($"{businessUnitName},{businessUnitToCompanyMapping[businessUnitName]}");
// Only add the update request when applying for real
if (!isSimulate)
{
multipleRequest.Requests.Add(updateRequest);
}
rowsAddedToBatch++;
Console.Write(".");
}
totalRecordsProcessed += rowsAddedToBatch;
if (rowsAddedToBatch > 0 && !isSimulate)
{
Console.Write("Sending {0} updates in a batch", rowsAddedToBatch);
var updateResult = orgService.Execute(multipleRequest) as ExecuteMultipleResponse;
moreRecordsExist = true;
Console.WriteLine(" done");
// If any faults are encountered, flag those IDs to not be processed again
// in subsequent batches.
if (updateResult.IsFaulted)
{
foreach (var response in updateResult.Responses)
{
if (response.Fault != null)
{
Console.WriteLine(response.Fault);
faultedIds.Add(((UpdateRequest)multipleRequest.Requests[response.RequestIndex]).Target.Id);
if (faultedIds.Count > 100)
{
throw new ApplicationException("Excessive number of update failures, aborting operation");
}
}
}
}
}
else
{
Console.WriteLine("No {0} rows remain to be updated for {1}->{2}", entityName, businessUnitName, businessUnitToCompanyMapping[businessUnitName]);
}
}
}
simulationWriter.Close();
simulationWriter = null;
stopwatch.Stop();
Console.WriteLine("Processed {0} rows for the {1} entity in {2}ms.", totalRecordsProcessed, entityName, stopwatch.ElapsedMilliseconds);
return (faultedIds.Count == 0);
}
/// <summary>
/// Gets an entity reference to the company with the specified ID if one exists.
/// </summary>
/// <param name="orgService">The CDS connection.</param>
/// <param name="companyId">The company ID to search for.</param>
/// <returns>An entity reference if one exists; otherwise, null.</returns>
private static EntityReference GetCompanyReference(IOrganizationService orgService, string companyId)
{
if (cachedCompanyReferences.ContainsKey(companyId))
{
return cachedCompanyReferences[companyId];
}
QueryExpression query = new QueryExpression("cdm_company");
query.ColumnSet.AddColumns("cdm_companyid");
query.Criteria.AddCondition("cdm_companycode", ConditionOperator.Equal, companyId);
query.TopCount = 1;
EntityCollection result = orgService.RetrieveMultiple(query);
EntityReference entityRef = null;
foreach (var entity in result.Entities)
{
entityRef = entity.ToEntityReference();
break;
}
cachedCompanyReferences[companyId] = entityRef;
return entityRef;
}
}
}