I worked to update our solution, I used the approch describe on ANZ giude Version Updating.
Following this apporch (it's very powefull) I'm able to upgrade my solution but I think it still remain an open point about migration.
On EF pre NetCore you can applied migration created in the past but now EFCore create a file to take a snapshot of db schema so is not possibile to add previous migration. We found 2 solution:
Is it possibile to use different approch to update db?
We had similar issue, in the past, ANZ solution < 6 related to geographical distribution of UI/SQL/REDIS. We enabled REDIS and put REDIS/UI in same region to speed up the query and latency disappear.
I hope this can be usefull to find your issue
I migrate from ABP 4.8.1 -> 5.13 and I found an issue on Swagger
if (WebConsts.SwaggerUiEnabled)
{
//Swagger - Enable this line and the related lines in Configure method to enable swagger UI
services.AddSwaggerGen(options =>
{
options.SwaggerDoc("v1", new OpenApiInfo() { Title = " API", Version = "v1" });
options.DocInclusionPredicate((docName, description) => true);
options.ParameterFilter<SwaggerEnumParameterFilter>();
options.SchemaFilter<SwaggerEnumSchemaFilter>();
options.OperationFilter<SwaggerOperationIdFilter>();
//options.OperationFilter<SwaggerOperationFilter>();
options.CustomDefaultSchemaIdSelector();
options.CustomSchemaIds(o => o.FullName);
options.OrderActionsBy(o => o.RelativePath);
}).AddSwaggerGenNewtonsoftSupport();
}
my DTO class is
public class InputDto: ...
{
public string FilterText { get; set; }
public long? Id { get; set; }
...
public OrderStatus? Status { get; set; }
...
public OrderStatus[] StatusList { get; set; }
...
}
The only way to fix is add a try catch
private static void AddEnumSpec(Type type, ParameterFilterContext context)
{
var schema = context.SchemaRepository.Schemas.GetOrAdd($"#/definitions/{type.Name}", () =>
context.SchemaGenerator.GenerateSchema(type, context.SchemaRepository)
);
if (schema.Reference == null || !type.IsEnum)
{
return;
}
var enumNames = new OpenApiArray();
enumNames.AddRange(Enum.GetNames(type).Select(_ => new OpenApiString(_)));
try
{
schema.Extensions.Add("x-enumNames", enumNames);
}
catch (Exception ex)
{
Console.Write("Ingore this error");
}
}
Hi @ismcagdas
I think we/you can add a warning on DbMigrator project when table will be delete to advice user.
I move my app to 5.11 and I find this migration
In the same bulk of migration I see a second migration
I think is so dangerous. In my case table is empty and no data lost but if I move from different version I can potential lost all data in this table.
I see that this migratio is only for raname colum why you decide to release this script and not a Add temp column copy data and remove column?
Hi @ismcagdas
yes it's correct. Consider that our solution (now has more or less 96 project). On this large solution we have 6 "service" already defined with their own DBContext and table so we have (core,app,shared,web for each service). We are moving to ABP.IO and to do that our path could be move one "service" into a ABP.IO microservice (we no that have some limitation (es. we can use only sync method beacuse RabbitMQ is not available on ANZ)).
Like has you suggest out path is:
So our open point still:
For this reason I've create this 2 empty project to create an internla POC a prepare the guide line to implement our pattern.
Our goal could be deploy different service / UI that it use ANZ for authentication (more or less that you do on public web site, support and commercial on ABP.IO) mixing ANZ and ABP.IO.
So now we can share access from ANZ & ABP.IO (at the moment only via authentication_code in next day we check different auth mode).
Now wich is the best approce to share "data"
ANZ is monolityc ABP is modular supose we develop a now module Product (or move some API from ANZ to ABP) respect the picture above we can't have GW.(our UI in Angulr)
How can we use data from new ABP service? From ANZ of course we can use normal TS proxy.
Hi I see that is a cache related issue. If I give permission to a scope client and not delete / refrash the Redis permission not works fine but I need to delete only key related to client and not all db.
Hi
ok I can understand it but why if authenticate my abp.io client on ANZ I get an 403 error and if I put a row on abppermissiongrant I get an invalid_grant.
ANZ only give ok or KO for user access (evenutaly can share some date on OpenId configuration like name,email, etc). The scope on ABP.IO give the role/persmission configure.
Are you able to make it work with my two solution? Colud you please share what we need to change?
Hi
has you see the 2 solution sended via wetransfer?