Yes, we haven't customized the login page, so it is a weird situation. Anyway, closing the ticket since we got it working. Thank you
I got this fixed by installing the old Abp 5 version back, saving the EnableLocalLogin setting as true from the Account settings and re-installing the new version. I think that I was not able to refresh the cache after all. I guess the Abp 7 (or 6) version contained a bug fix that now the setting is correctly considered in the login page? However, I noticed that one of our tenants also had that setting as False, and their login page did not give that error. There might still be the same bug in the tenant side?
Hey, we have a pretty urgent issue where we cannot login into our host side in one of our production environments after Abp 7 update. For some reason the login page now shows this for host side login. Tenant side logins work normally
Can you give us pointers on what affects this. I know there is a Abp.Account.EnableLocalLogin -setting that for some reason was in False for our host, but I think that value has not changed in the Abp 7 update. I switched to True in database and tried to refresh the cache but that did not fix the issue.
Yes, that seems to work. I got the impression that that was the fix for the hanging tests, but everything seems to work without it. Thank you for your help!
Ok, thanks. Do you have any suggestions on how to handle our current problem? We didn't have nested transactions problem before, but now we do. What caused the change? Why does the nested transactions problem occur during the tests? As far as I know, Abp unit of work system should be able to recognize if there is a transaction open already.
Is this a new limitation? This worked fine earlier.
Ok. that " context.Services.AddAlwaysDisableUnitOfWorkTransaction();" actually brought another problem for us. With some of our integration tests, we use test collection to speed up the test execution. We used the transactional unit of work to control the system state, so the tests were not dependent on each other. System state changes were rollbacked after the test. Now that the transactions are disabled, these tests are now interdepended and some query tests fail because another test has affected the database state. I know that Abp recommends to do the test initilization per test, but with larger test sets, that makes the test excecution slower and slower. Is there any way to control the transaction separately per test now? Is there any other good way to speed up the test excecution with larger amount of integration tests?
hi, that seemed to fix the problem.Awesome, Thank you! Can you please explain why that helps, so I can put a comment in our code?
We have a big problem with our backend unit tests after we migrated from the 5.3.1 version to the most current (7.0.2) version. Now some of the tests are hanging in execution indefinitely when running large batches of tests at once. The same thing happens whether we run the tests from VS or through the "dotnet test" via cmd. From my investigation, this might be due to using synchronization with async-methods.https://github.com/abpframework/abp/issues/2075#issuecomment-561582158. One obvious place where this is done is the common data seeding in the test base.
private static void SeedTestData(ApplicationInitializationContext context)
{
if (context == null)
{
throw new ArgumentNullException(nameof(context));
}
AsyncHelper.RunSync(async () =>
{
using (var scope = context.ServiceProvider.CreateScope())
{
await scope.ServiceProvider
.GetRequiredService<IDataSeeder>()
.SeedAsync();
}
});
}
For validation, I created a structure that does the seeding through IAsyncLifetime initialization so the seeding is done properly through the asynchronous call stack. That seemed to mitigate the problem alot but there were still some tests that were left hanging. After that, I disabled the parallel test execution. With these changes, I can get larger batches of tests to finish execution. Forcing non-parallel execution is not ideal, however. Have you seen this problem occur previously? And what can be done to fix the issue? I checked that, we don't use synchronous calls for async methods in our own code.