What permission do I need to assign managed identities to app roles? - azure-logic-apps

I want to authorize several Logic Apps to access operations on an API secured by an app registration, which has several app roles describing different operations. Currently a directory administrator is doing this manually using New-AzureADServiceAppRoleAssignment once the Logic Apps are created, because the associated service principal doesn't exist until then.
I'd rather this were automated, because especially in development, the manual work of asking a directory administrator to re-run this script is very tedious. However I don't know how to grant the scripting account - a service principal linked to a DevOps service connection - only the permissions to do this, and not make it a directory administrator able to do anything. If the service connection is a directory administrator, developers would be able to supply it scripts to tell it to create or delete any combination of role assignments, enabling them to let themselves into anything, and also shut out people who should be able to prevent this. This is an unacceptable security hole.
What is the minimal permission needed to allow an automated process to script the creation of app role assignments but nothing else, and where is this documented?

For this requirement, you can just add the scripting account as owner of your app(just create a new user without any roles and add it as the app's owner). Then it can just add role assignment in this app but can't do other operations on role assignment of other apps.

Related

Insufficient privilege for operation like Get-AzureADApplication and Set-AzureADApplication

I would like to automate deployment and it requires to update settings for Azure AD Application registration.
So far I am able to :
create an Azure AD Appregistration and Service Principal with certificate (thx MS documentation)
then use command Connect-AzureAD with previous service Principal with its certificate
use command like Get-AzureADApplication -ObjectId 11111111-2222-3333-4444-555555555555
In previous bullet ObjectId 11111111-2222-3333-4444-555555555555 match with application i created on first bullet
However i am unable to execute command like:
Get-AzureADApplication -Filter "DisplayName eq '$aADApplicationame'"
and $aADApplicationame matches with application created previously
Set-AzureADApplication -ObjectId $aADApplication.ObjectId -ReplyUrls $ReplyUrls
Get-AzADServicePrincipal
I get following error message
Set-AzureADApplication : Error occurred while executing SetApplication
Code: Authorization_RequestDenied
Message: Insufficient privileges to complete the operation
Based on my research, i set up some API permissions as follow:
Unfortunately no luck and still get insufficient privilege although all permissions were granted.
Do you know if I miss something ? Is there any specific permissions i should add to make it works ?
Regards.
As mentioned by another reply, you could give the Global Administrator role to the service principal, it is correct, but the permission of Global Administrator is too large in this case, it may cause some security issues.
In this case, the commands Get-AzureADApplication and Set-AzureADApplication you used essentially call the Azure AD Graph API, so to solve the issue, a better solution is to add the permission of Azure AD Graph API, please follow the steps below.
1.Navigate to the API permissions of your AD App -> select Azure Active Directory Graph(not Microsoft Graph).
2.Select Application permissions(not Delegated permissions) -> Application.ReadWrite.All -> click Add permissions.
3.At last, click the Grant admin consent for xxx button.
After a while, try the commands again, it will work fine.
Update:
After I check the doc, I find there are already some new commands released by MS which call the Microsoft Graph, haven't seen them before.
e.g. In your case, you can use Get-AzureADMSApplication instead of Get-AzureADApplication.
Get-AzureADMSApplication -Filter "DisplayName eq 'joyttt'"
Use Set-AzureADMSApplication instead of Set-AzureADApplication.
Set-AzureADMSApplication -ObjectId <object-id> -Web #{ RedirectUris = "https://mynewapp.contoso.com/" }
For Get-AzADServicePrincipal, there is no equivalent currently, there should be one in the future. When using the commands above, the permissions of Microsoft Graph will work, no need to use Azure AD Graph, but you should use Application permission, not Delegated permission (you used the Delegated permission in your question.)
You are facing this issue because Powershell cmdlet works differently than compared to MS Graph. Instead of permissions, Powershell require roles to do this operations. Please add Global Administrator role to your service principle and then try the Connect-AzureAD so that, the issue will be fixed.
For more details, you may refer to Assigning administrator roles in Azure Active Directory.
I was also facing similar issue, make sure are doing below two things:
Set Run as account for azure automation account
In newly created app registration for azure automation account after setting Run as account, make sure you add Cloud application administrator role explicitly.
Add API permission for Application.ReadWrite.All (Microsoft graph)
In my case the app registration was showing cloud application administrator role under Roles and Administrator screen, which I thought gives the new app registration required permission but that was not the case. PowerShell script only worked after we assigned the cloud application administrator role explicitly.
In the beginning thanks for previous posts it gave a lot of inspiration according topic. Problem occurred in our case at automated bicep mechanism that is supposed to add API permissions for Microsoft Graph.
Error: Authorization_RequestDenied
Solution:
We needed to give Enterprise Application running mechanism Microsoft Graph (not Azure Active Directory Graph it will be deprecated) Application permissions:
Application.ReadWrite.All
AppRoleAssignment.ReadWrite.All
Directory.ReadWrite.All

What are good practices for granting database permissions to a web service connection?

I have been searching for articles and SQL script examples that would demonstrate how to securely and conveniently solve one of the most common scenarios - connecting from a .Net Core Entity Framework based web application to an SQL database.
But somehow I could not find any coherent step-by-step guide from a reputable source.
Let's assume the following:
I cannot use integrated Windows auth in the connection string and must use username and password based auth (because hosting on a Linux server and the DB is on a different Windows server)
the web service will need your usual minimum set of permissions - connect to the database, read data, write data, delete data, execute stored procedures
While reading many tutorials, I find there are multiple ways to manage the connection permissions. To avoid this question being too broad, I'll list my current choices as I understand them (please correct me if I'm missing something).
Users and logins:
create a login and a user for the database
create a database-only user without a login (not sure if this is applicable to a web app and connection string, but still it's a feature that I've seen being used)
Assigning permissions:
assign the user to some fixed SQL role (db_datareader, db_datawriter AND also will have to grant EXECUTE permission)
grant all fixed permissions
create a custom role (let's say, db_web_apps) with required permissions
Which choices are better (more secure and easier to manage in general) and recommended by SQL DBAs?
I think every database administrator should have a template script handy for quickly adding a new user with minimum required permissions every time when developers ask for a new connection for their shiny new web app.
If you know a good, reliable tutorial or GitHub / Gist example that explains what and why is being done that way or a script that you yourself have used for years without any issues in production environments, I'll really appreciate if you could share it.
Create a role in the database and assign the required privileges to the role. Don't use the fixed database roles. Instead grant permissions directly to objects, schemas, or the entire database if necessary. Like this:
create role trusted_app_role
grant select, insert, update, delete, execute
on schema::dbo to trusted_app_role
That will grant the role full DML permissions on all the objects in the default dbo schema. So if you have any tables or procedures you don't want the app to have access to, just create them in a different schema, say, admin. This way you never have to fiddle with permissions as you add objects. The fixed database roles predate schema-based permissions, and aren't really needed any more.
For your application's identity, add Active Directory or Azure Active Directory (Azure SQL) identities to this role, or, if you can't, add SQL Users to the role.
If you are on Azure SQL, you should normally use a database user without a login. On SQL Server you can only add "contained database users" if you enable Partial Database Containment. Which you can do, but is incompatible with Change Tracking and Change Data Capture, so it's a tradeoff.
So normally for SQL Server you still create a login and map the user to the login. EG:
create login web_service_user with password = '5X+jeuAB6kmhw85R/AxAg'
create user web_service_user for login web_service_user
And then add that user to your role
alter role trusted_app_role add member web_service_user

assign permission to group managed service account

I have been advised that it is better to run a scheduled task as a Group Managed Service Account (gMSA) rather than as a domain user account. I can find plenty of information about how to create the gMSA, and how to configure the scheduled task to run as that gMSA, but all of the tutorials and training I have found stop there. I can't find any information on how to assign permission to that gMSA.
For example, a scheduled task needs permission to write to a folder. Normally, if the scheduled task is running as a domain user, I would go into the properties for that folder and set the security so that the domain user has write permission. But I am finding that I cannot give a gMSA permission in the same way.
What am I misunderstanding about gMSAs?
I have not find a way to assign permissions to a GMSA directly to the file system. My workaround is to add the GMSA account to an AD group and then assign permissions to the group.

How to securely store my CouchDB admin password?

I spent a long time yesterday to configure for my CouchDB instance in order to create a little app and letting CouchDB manage authentication and authorizations for me.
So I ended up with something like that :
On top of everything I've got a server admin, who basically is god on my CouchBD instance.
Then I created a database named "mydatabase" (for example) and added the role "mydatabase_dba" as admin and also the role "mydatabase_user" as reader.
I also created a database named "_users" which contains all the database admins and users with their roles and also a design document named "_auth" which manages authorizations.
Only the server admin is admin of this database, and I added users with role "mydatabase_dba" as readers. Then, for those of you who knows about it, I modified the "validate_doc_update" field o the "_auth" document so that users with role "mydatabase_dba" can only deals with users with role "mydatabase_user".
So, to summarize at this point :
server admin is still god
users with role "mydatabase_user" can connect to "mydatabase" but they are just readers
users with role "mydatabase_dba" are admins of "mydatabase"
users with role "mydatabase_dba" can connect to database "_users" where they are readers
users with role "mydatabase_dba" can only manage users of role "mydatabase_user" in "_users"
Hope this is clear :D
What I can do now is create an application that will not manage users itself, but let users connect to CouchDB directly (transparently).
The problem come when it deals with users creation/update/deletion.
Because only users with role "mydatabase_dba" can access to the "_users" database and work on users with roles "mydatabase_user", I need at some point to connect to CouchDB as this db admin.
I have two solutions :
Create a user interface into my app that will let the admin connect and do what he has to do
or
Make some more code and let the app do it automatically, this is the solution I prefer, but the problem is : I have to store the admin credentials...
Sorry for the long introduction but I had to describe the landscape first :)
I created a post yesterday about how I could secure the connection between my app and the CouchDB instance : here
The solution I was given is to use HTTP over SSL (/TLS) to secure the communication. I'm okay with that, but now I have another concern, maybe I'm paranoid, but because my app will need to connect as "mydatabase_dba", I have to store its credential somewhere.
But how to store them securely ? As said in my previous post, even if I store the hashed password instead of the plain text password, if an attacker access my app source code, he'll have my admin credentials...
An application should never have an administrative rights. It should only be given the bare minim rights it needs to function. If the application needs some administrative rights, make sure it has as few as possible. Other than that, most of the time these credentials are stored in plain text in some file that only your application can access.
Never commit this text file into your source code manager (Subversion, Git, etc.)! Placing the file into a running system must be a step in the installation procedure.

Is it true that SQL auth is only great for multiple role apps?

I believe Windows auth is the best practice to use to connect to SQL DB. I am hear talking about application user account..
Is it true that SQL auth is only great for multiple role apps and window auth is only good for single role app? I never heard that windows auth with muitple role os only good for smaill internal app?
multiple Windows logins = multiple connections = no pooling = poor scaling?
The problem with using Windows auth for a web application is that many web applications store their application users' credentials in the same SQL database that is used for other application data.
So you have a chicken-and-egg problem. You can't authenticate the user before connecting to the database, and you can't connect to the database without authenticating the user.
It should be possible to use Windows authentication, and then also have application-specific attributes of the user stored inside the database. But most people find this cumbersome to administer, and also limiting to portability of the application.
For example, if one of the feature of the application allows users to change their own password, then the process running your web application needs the privilege to alter a Windows password, which may mean that the application needs to run with Administrator privileges.
If you let the application manage user ID for the context of the application, then to change a user's password is just an SQL operation, and your application is in charge of enforcing security for that.
I'm not sure what you mean by single-role and multi-role app. I have built apps before where there are multiple SQL Server Database Roles, each with a Windows Domain Group of users allowed in that role. So user management is completely within Active Directory, with a 1-1 correspondence between the Domain Group and the Database Role.
We typically did not manage the security within the application itself except obviously declaratively during the database creation where each object was granted access by particular roles according to the design. Typically, in a simple case, we relied on db_datareader role being granted for general usage to non-specific groups of users like database and network administrators for troubleshooting or report-writers or business analysts for ad hoc reporting. Actual users of the app would be granted execute on the relevant SPs to be able to modify any data (so all data creation or modification was through SPs and only explicit members of the ThisAppsUsers AD group could do it). Any advanced SPs (say, merging or deleting accounts) would only be accessible by ThisAppsAdmins AD group. And that was usually all we needed for moderate-sized applications. For more complex functionality, it was also possible to interrogate AD directly for custom attributes (user is an admin only for this customer account but for others is just a user)
This same technique can be used with SQL Server logins, but of course the individual SQL Server logins have to be added to the database roles, and you don't have the richness of AD and have to build some kind of directory service into your database.
The ability to even use AD may not be possible for many applications, so in that case, the security architecture would obviously have to cater to that model.
using the integratedSecurity=true option for SQL JDBC , by including the JDBC auth .dll, should give you database connectivity without authenticating...

Resources