First, enable a Managed Identity for your Logic App:
In the KeyVault, add a new Access Policy for the new Managed Identity (from the previous step). Use the least priviliges. In my case it is just enough with GET for secrets.
Instead of filtering values from an Azure Storage Table, you can do it directly in the bindings. It might not be a solution for everything, but in the right place, it is fantastic. I was very surprised to see how little code was needed after this binding change:
For that to work, define the filter attribute in the bindings: “filter”: “(PartitionKey eq ‘{package}’)”
To try it out, add a new row in a table defined in the bindings (“metadata” in my case):
Start the function app and navigate to your function:
Just a quick tip today. I hope this helps you in your work. The raw material comes from stackoverflow:
Site Collection App Catalogs are great for special cases (like developing apps or site unique apps), but using them on scale would be a mess.
I got a question: Why should we use the Tenant App Catalog at all when we could enable a Site Collection App Catalog on every teamsite? So the suggestion here is to install SharePoint Framework Packages on many Site Collection App Catalogs, instead of the Tenant App Catalog. In that way those wouldn’t be visible for all users in the “Add an app” page.
Technically, both ways are possible, but I would say, we should use the Tenant App Catalog for apps that are installed on many sites. I have gathered some pros and cons to help people decide.
But first, some abbreviations and notes on terminology:
SCAC = Site Collection App Catalog, (Local App Catalog)
TAC = Tenant App Catalog, (Global App Catalog)
App = SharePoint Framework Package, but also a SharePoint Add-In
Pros and Cons of Site Collection App Catalogs
👍 A SCAC makes it possible to have local instances of an app. It is perfect if you want to test an app without making it visible for the whole tenant. It is also good for apps that are only installed on one site collection. There are benefits of distributing site specific apps through the TAC, such as better governance, once installed on a site, those apps can be Disabled.
👎 Mess in all the apps and the versions. There might be different versions of the apps. The update procedure will require a special script to find all the sites and run the update. This also requires owner access to that site. If the owners would like to remove all the IT accounts in order to collaborate on confidential information, then it won’t be possible to update that site. It will lead to poor governance. Just to give it a perspective. Imagine that every smartphone had its own local App Store/Google Play. Possible but counterproductive.
👎 A SCAC is a security risk. The control and governance is handed over to site owners. It means that they can install any apps on their site collections. Poorly designed and implemented apps can lead to big security issues, in the similar way as custom script.
👎 Enabling a SCAC adds additional lists and features (such as Apps for SharePoint). It will require more space (not so much, but still), but first of all it will make some users more confused, actually more confused then some more available apps. Imagine all the support: What is it, Oops I deleted “Apps for SharePoint”.
👎 Every app will install the resources on every site collection, meaning every app will have multiple copies of the same app. It means more space needed. That is of course, not the case if you use Azure CDN.
👎 Every app will use its own urls to resources, preventing browsers from gaining performance by caching the resources cross site collections.
👎 If Azure CDN is used, you must make sure all the versions of javascript and css files are kept because you might not know which versions of an app are still used.
Pros and Cons of the Tenant App Catalog
👍 Easier to control the apps and have better governance. The Tenant App catalog tells what apps exist in your tenant. As admin you can verify that the code complies with the guidelines (by running code checks), you can make sure, every app there has all needed information such as app owner, description etc.
👍 Easier to update apps. Installing a new version of an app will update all the instances. There won’t be any custom script to traverse all the sites with that app.
👍 It is more cost efficient to follow the Tenant App Catalog approach, because it is the recommended way of installing and updating apps. Having scripts and procedures for SCAC, (even completely automated) will always cost more: to keep the scripts up to date, support etc.
👍 Benefits of the common apps. If a business unit comes up with a good solution, it can be used by other Business units, if it is discoverable and follows quality guidelines (more on that later).
👎 The Apps from the TAC are visible to all Business Units and the users might be confused. This issue can be mitigated in two ways:
The business units that have not rolled out SharePoint Online and don’t use any apps, can remove their users from the App Catalog.
Having a good naming policy, app icons, good descriptions, disabling any “auxiliary” apps. On top of that, some user guidance.
Summary of pros and cons
To summarize, Site Collection App Catalogs for reusable apps installed to many site collections, is a bad idea, because it means a security risk, less control and governance, encreased data and traffic to SharePoint Servers.
How can we scope apps to our business units
Are there other ways of scoping the apps to business units other than misusing the Site Collection App Catalogs?
First of all, it is a question of user adoption. It is better to help users instead of breaking the functionality or misusing other concepts.
Second, it is a question about governance of the Tenant App Catalog. It should be clear how an app can be added, and how that package should be wrapped:
good naming standard,
proper localization
a mandatory icon,
good description,
owner/contact person. Also, only relevant apps should be “Enabled”. All the other apps should be Disabled.
A couple of apps from your organization is a drop in the ocean compared to SharePoint Store Apps
Control with Permissions
If you remove a user from the Tenant App Catalog, the apps will disappear for that user.
There are apps, but this user cannot see any apps to install.
The permissions can set on the whole Tenant App Catalog by removing “Everyone except external users” and then adding all AD Groups who should see the apps.
The permissions could also be set on the specific packages or folders.
The permission way of scoping the apps might work. But it is important to know that it is a workaround and it breaks the default setup (where all users have READ on the TAC). This means that along the way you can bump into limitations or bugs, more or less unique to your organization. It will be harder to get help from Microsoft or from community.
I uploaded my app to two site collection app catalogs. Do I need to get approval for the app twice? – No.
I have got my approval for Delegated Groups.ReadWrite.All for App X. Now I want to use the same permissions in another app, do I need a new approval? – No.
The steps described in these guides are easy, but that effort made me think about the first pair of pros and cons.
A pipeline variable is faster to configure
A variable in a pipeline takes zero time to set up. Also, A secret variable remains as a secret, since no one can read it in plain text. To configure the Key Vault way of getting secrets requires admin time. Unless you have Admin rights in your Azure Active Directory and your Azure Subscription, you might need to request and argument for one or more of the following:
A service principal (an App registration) with a secret.
An Azure Key Vault (and maybe a resource group) with an Access Policy for the service principal
A service connection in your Azure DevOps Project
Of course, most of it is one-time-job. But still, in many organizations it will require good preparation. The pros for an Azure Key Vault secrets in a pipeline is that
Admins can manage the secrets centrally from Azure
Set it up once and let Azure DevOps people use it and re-use it in many pipelines, but still you need to set up a new Service Connection in every Azure DevOps Project
The fact that it is easier to reuse lets me think of my second pair of pros and cons.
A pipeline secret variable is more secure
Let’s say you need a password to a service account that will upload something important, e.g. an account that will upload a new SPFx package to the SharePoint App Catalog.
Doing it the pipeline variable way means that it remains as a secret on that particular release pipeline. Only release administrators of that project can alter the pipeline steps. No one else.
Doing it the Key Vault way, means that you must watch out on every part of that chain:
Users who have access to the Key Vault in Azure.
Service Principal that can read the secrets through access policy. Who has access to the secret?
Service Connection in an Azure DevOps Project. Who can use this Service Connection – to add and modify release pipelines? By default, all Release Administrators can do that. To do it more secure, you need to limit the count of Release Administrators. But it means less flexibility in a team and more admin effort for the allowed Release Administrators.
Also, the service principal used for getting the secrets and in the service connection, should not be reused across projects in Azure DevOps. Dedicated Service Principals will make it more secure because misuse can be more easily discovered and stopped – and thats on a project level, not for all service connections.
Summary
In flat small organizations, using a Key Vault for using secrets in Azure DevOps Pipelines is great, it saves you time. But it is less secure, and requires time and effort for an appropriate security.
I got my fingers and tried that guide out. This worked good, I spent some time, though, googling (binging) around to get rid of the SSL Warnings for the remote “localhost” on my Mac.
I would like to share this simple instruction on how to trust a self signed certificate from gulp-connect on Mac OS. The implication is that the certificate is on the remote linux machine (on the Visual Studio Environment), that you are connected to through the Visual Studio Code extension.
The first step (after you have connected and set up a project) is to download the certificate. It can be found in the following directory:
Choose a folder (like Desktop or whatever) to save it to. Then double click server.crt to open it in the Keychain Access.
In the Keychain Access, locate the certificate, it will have the name “gulp-connect”. Open it and enter the “Trust” section. Under “When using this certificate” select “Always Trust”.
Allowing custom script has its security implications. But what exactly does it mean? Is it dangerous? My colleauge Daniel and me have done a little experiment. There are two implications stated on MS Docs:
Scripts have access to everything the user has access to.
Scripts can access content across several Office 365 services and even beyond with Microsoft Graph integration.
To summarize, we can look at that picture:
So the risk that user 1 (the Blue User) intentionally or unintentionally places a script and lets user 2 (the Red User) run this script by linking to the page that has this script. The page must be in a “common” place.
Let’s try that. Let’s have a little hackathon.
Experiment A. Very Simple.
Scenario: the Red User (my colleague Daniel) has a site with with sensitive information (He has a list called Secrets). The Blue user (me) does not have access to that site. The Red User accidentially shares the link to the secret list. Now, to prove the security risks with the custom script, the Blue User has to get the secret information.
To make it more visual and fun, we have found a box where we have put candies and locked it with a combination lock. The code (the combination) is stored in the Secrets list.
The combination consists of three digits from 0 to 9. So there are 1000 possible combinations.
Since the Red User knows that he is going to be hacked, he acts as normal as possible, not especially suspicious. And since the experiment A is a simple one, we let the Red User to accidentially show the link to the Secrets list. That can be:
A screenshot that shows something else but contains the confidential site url
An email that is sent to a wrong recipient
A pasted link in a wrong Teams chat
etc.
So the Blue User finds out that there is a Secrets list that has the following address:
/sites/DailyWork/Lists/Secrets
“Fortunately”, the Blue User has access to a classic site (where Custom Script is allowed). He shares that site with the Red User. Now both the Blue User and the Red User have it in common.
Then, a new Web Part Page is added with a tiny script.
That script makes a REST call to the Secrets list and uploads the result in Shared Documents. For the sake of simplicity, a built-in document library is used. An attacker would probably add a hidden list for that.
Now to get the Red User to that page, the Blue User adds some “important” information so that the Red User will not become suspiciuos.
The “important” information and the actual script editor
Now it is just the matter of sending a link to the Red User. The Red User navigates to that page and the script is executed while he is reading the information on that page.
The Blue User gets the Secret that was generated by the Red User:
With that information the Blue User has hacked the combination lock and he can now open the red box with candies!
Skitgott Candy
Experiment B. Harder
For now I think it is enough with the Experiment A, the Simple Experiment. We have proved, that, indeed, security is compromised with custom script allowed. A user can get confidential information from another user. Maybe some of you want to set up a more sophisticated experiment. Make sure the “hacked” person is warned and agrees to be part of an ethical hacking act. We maybe can try more to show the importance of keeping custom script disabled and not using classic sites that have it by default.
Wrapping up
Custom Script (or DenyAddAnCustomizePages=false) is dangerous, it is of course not black and white. It provide the ability to have small appications directly on sites created by users. But as with all other aspects of IT, you should know what you are doing. Every site with custom script is a potential trojan horse. Minimizing the number of classic sites and sites with custom scripts lowers the risk. All the collaboration sites should be modern team sites without custom script.
Only the server relative url part is counted. The “https://%5Btenant%5D.sharepoin.com” has no impact on that. It starts from the foward slash: “/sites” (or /teams).
An encoded value like blankspace (%20) is treated as one character, not three.
A unicode character, and an emoji is treated as one character. Good news for Non-English Names.
Url Parameters, like “?Web=1” are not calculated.
The site url and the document library url is taken into account
All slashes are included
A file extension is also included, and even the dot, e.g. “.docx”
Other related information
A site url and a group name can only be 64 characters max.
The path in the “Copy Link” is much shorter than the “real” path
There is no limit (as of time of writing – 2019-10-30) on the folder name length (other than the bigger limit of 400 characters), I had no issues to add a folder name with 312 characters.
Those are parts that are counted in the Path Length
Recommendations for Folder-heavy document management
I don’t want to discuss whether to folder not to folder. On that topic, my favorite is the slide deck with the same name by Bobby Chang: To Folder or Not To Folder. For those who need to use folders I would recommend:
Try to have a short site url/group name
Try to have a short document library url. Why not creating just “docs” instead of “Our very important documents”? Note, that I am talking about the url, not the display name. The trick is to call it “docs” (or some other short word) initially (which will turn to the url), and then you can name it to whatever you please.
Even if you use folders, try to flatten the structure.
I saw a demo of it on the European SharePoint Conference in Copenhagen in 2018. Sebastian Fouillade, who showed this, compared this big change with brain surgery. All the urls, all the connections. But now it is possible. Today I have seen it even in my standard release tenant.
It is really appreciated. Soon it will be possible to rename misspelled sites, like “devlepment” to “development” etc.
To rename you need to go the SharePoint Admin, find your site among Active sites and click on “Edit” in the site url area.
I also can image, it will be very handy to change the url of a SharePoint site that was automatically created for a Team (through the Office 365 Group). The team might have some longer name, but a simpler url is often appreciated.
I have tried and seen that also the automatic redirects from an old site url to a new site url works.
Caveats and Limitations
mailNickname ≠ site url
Now it is even more important to not to rely on the fact that mailNickname of an Office 365 Group and Site url are the same. As Elio Struyf describes, it is not a good idea to compose a URL from the group name. I have used in PoCs the site url to get the group id:
Now it the amount of cases where it will work, will be less.
Not all site types can change url
I have found that the App Catalog site cannot get a new url, neither sites with Publishing Features.
The official message from Microsoft
We’re making it possible for SharePoint administrators to change site URLs.
We’ll be gradually rolling this out to customers in mid-October.
The roll out will be completed worldwide by the end of October.
This roll out does not apply to Office 365 subscriptions for EDU tenants.
This message is associated with Microsoft 365 Roadmap ID 56205.How does this affect me?
As a global or SharePoint admin in your organization, you will be able to change the URL for the following types of sites (previously called “site collections”):
Classic team sites
Communication sites
Modern team sites that don’t belong to an Office 365 group
Office 365 group-connected team sites
For example, if you have a site named https://contoso.sharepoint.com/sites/Develpment, you can rename the site to correct the incorrect spelling of “development” via the SharePoint admin center.
Automatically-generated redirects will ensure that old links do not break.What do I need to do to prepare for this change?
There is nothing you need to do to prepare for this change, but you may consider updating your user training and notifying your help desk.
A non-admin user can create no more than 250 resources in Azure AD. That is one of the many Azure AD service limits and restrictions. A “resource” can be an app registration, an Office 365 Group etc. But I would like to discuss Groups more in detail.
Imagine the following scenario: Your organization has disabled Office 365 Group Creation. Only IT can create new groups. A service account has been set up for creation of team sites. The application permissions are “binary”, either everything or nothing: Group.ReadWrite.All. This service account will hit the limit very soon.
To prove that, I have created a small script that creates 251 groups.
By the way, just for clarification, when create a new group, that will also create a SharePoint site.
Please don’t try this with your real account in production. The 251st request will fail:
Directory_QuotaExceeded
The directory object quota limit for the Principal has been exceeded. Please ask your ad ministrator to increase the quota limit or delete objects to reduce the used quota.
Even if you remove, it will take time to get free slots in this limit:
Deleted Azure AD resources that are no longer available to restore count toward this quota at a value of one-quarter for 30 days.
There is not much to do about it. For App Registrations you can create and assign a custom role. But for groups there is no custom roles available.
Exceptions
It might be obvious, but still:
Admins
Admins do not have this limit. But not all “admin roles” are really admins. Those roles are excepted:
Global Admin
SharePoint Admin
…
Those roles are not excepted:
Message Center Reader
…
I don’t have time to try every admin role, but I suppose only admins that can change global configuration, are excepted, not the reader ones.
Communication Sites
Since communication sites do not have an Office 365 Group behind the scenes, a non-admin user will still be able to create such sites even after the limit is hit.
Workarounds and Solutions
Since my scenario for creating groups with a service account does not work, we need to seek workarounds and solutions.
Do not restrict Group Creation
That is the best one. If users can create groups/sites by themselves, then none of this would be a problem. But still, in my scenario, there is a business requirement to control the creation of groups.
Application Permissions Group.ReadWrite.All
That is exactly the opposite of my scenario. This gives that application full access to all groups and files (!). This means, that application can access all Group-Connected SharePoint Sites as well.
Microsoft creates permissions for groups
If we also had “groups” permissions for custom roles, then we could do the same way as with app registrations. Today (2019-10-25), there are only permissions for applications.
Microsoft creates new permission Group.Create.All
If there were a permission for only creating groups, that would solve the problem.
What we have today
There is a similar role: User.Invite.All, it allows only invitations, not editing All Users.
Microsoft allows exceptions per user
If there were a switch for the 250-limit per user, that would also solve the problem.
Granting the service account admin rights
Granting SharePoint Admin would solve the problem, but at what price? That is safer than Application Permissions Group.ReadWrite.All, since you need to actively add this account to the groups in order to read all the files, but this is still less secure than just a non-admin account.
Having multiple service accounts
If we had account 1..100 and we used every account 250 times. Theoretically it should work, but it is a cumbersome process. You need to keep track of how many groups an account as created, or having the right error handling. How should the password be kept safely. Should the accounts be removed when they have reached the 250 limit?
Group Creation Microservice
To overcome the limits and the ungranularity in the built-in permissions in Office 365, one way to solve it would be a tiny, but a dedicated, and secured service for creation of groups (and sites). It would still need the “hefty” Group.ReadWrite.All Application Permissions, but making it do the only thing and do it right, would mitigate the risks.
It could be a simple Azure Function that few have access to. That could be just a couple of lines of code.
Recent Comments