Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Urgent] Publishing Portal after new release breaks AAD #822

Closed
ZarTrox opened this issue Aug 10, 2020 · 49 comments · Fixed by #824
Closed

[Urgent] Publishing Portal after new release breaks AAD #822

ZarTrox opened this issue Aug 10, 2020 · 49 comments · Fixed by #824
Labels
P0 Critical issues.

Comments

@ZarTrox
Copy link

ZarTrox commented Aug 10, 2020

Bug description

If one publishes a DevPortal Instance after the latest release (07.08.20) log-in via the AAD button does not work.
The page loads but nothing happens at all. Also one cannot navigate through the URL.

Reproduction steps

  1. Go to the published version of a DevPortal Instance
  2. Get directed to the Sign-In page
  3. Press the AAD button
  4. Enter your credentials
  5. Wait
  6. See that nothing happens
  7. (If the Portal is resetted it is in a refresh loop)

Expected behavior

It should work like before the new release.

Is your portal managed or self-hosted?

Managed

Additional context

2020-08-10_12-59-07

@TimMnz09
Copy link

We have the same issue with our developer portal after publishing portal, but we are using simple username & password authentication.

@mikebudzynski mikebudzynski added the P0 Critical issues. label Aug 10, 2020
@mikebudzynski
Copy link
Contributor

mikebudzynski commented Aug 10, 2020

Thanks for reporting the issue, we are now investigating it.

@mikebudzynski
Copy link
Contributor

@ZarTrox we identified an issue with AAD: #824 but:

  1. We were able to reproduce it only in the self-hosted portals, not managed
  2. The website doesn't reload infinitely as you show in the screenshot

Can you make sure there weren't any changes in your AAD / networking setup? If there weren't any, can you provide us with the service name and network traces via email to apimportalfeedback@microsoft.com?

@TimMenze what's the problem with the username and password sign-in? Can you provide us with reproduction steps? It doesn't seem to be related.

cc: @ygrik

@TimMnz09
Copy link

@mikebudzynski sure, please find some more information below.

Reproduction steps

  1. Open Dev-Portal URL
  2. Redirect to Sign-In page
  3. Enter Username + Password
  4. Press "Sign In" Button
  5. Wait
  6. Infinite Refresh Loop to Sign-In page

login_problems_apim_developer_portal

@ygrik
Copy link
Collaborator

ygrik commented Aug 11, 2020

@TimMenze did you try to remove cookies and clear session?

@AnRei123
Copy link

AnRei123 commented Aug 11, 2020 via email

@TimMnz09
Copy link

@ygrik yes I did. Same problem :(

@MelHarbour
Copy link

We are also experiencing this issue with it looping after signin on AAD.

@LockTar
Copy link

LockTar commented Aug 12, 2020

Same problem here! No working developer portal anymore

@olandese
Copy link

we are experiencing the same problem. Not able to login with AAD anymore

@mikebudzynski
Copy link
Contributor

mikebudzynski commented Aug 12, 2020

We have hotfixed all regions, please republish your portal. Apologies for the inconvenience and disruption.

We'll create a new self-hosted release shortly.

@olandese
Copy link

I just did a republish, but the problem is still there (West-Europe)

@mikebudzynski mikebudzynski reopened this Aug 12, 2020
@mikebudzynski
Copy link
Contributor

mikebudzynski commented Aug 13, 2020

We redeployed the fix, can you please check again? Please, republish the portal and, if it still doesn't work, clear the cookies and the session.

@AnRei123
Copy link

Unfortunately, for the two affected instances (managed) it still does not work. I cleared the cache in the Chrome browser. Closed the browser and opened it again. Published the current version. Opened the web url in the incognito mode. But still same behavior. Only the sign-in page is shown after adding the credentials to the sign-in dialog. For the legacy portal of the impacted instances, this issue does not show up. The sign-in for the legacy portal with the aad button is possible w/o any problem. This issue only occurs for two of our instances for the new dev portal after I have done content modifications and after I have published the content. I also checked the behavior with the Firefox browser this morning. Same issue.

@olandese
Copy link

same here, I republished the portal, open it in an incognito window but the problem is still there

@butsona
Copy link

butsona commented Aug 13, 2020

We are also experiencing issues with sso. After clicking the AAD Signin the popup is shown. Once credentials are entered the popup disappears but the main form is not refresh. If we manually refresh the user is logged in. But users are unlikely to manually refresh.

Please can you make sure changes are tested before releasing. Especially when this effecting our clients.

@olandese
Copy link

olandese commented Aug 13, 2020

I was able to solve the problem disabling the "Redirect anonymous users to sign-in page".

image

If the option is enabled after logging in with AAD, the users seems still to be anonymous and no Products/APIs are shown.
With the option disabled (and after logging in with AAD) we are able to see the APIs and Products.
In our setup the Products and APIs are shown only if an authenticated user is in a specific APIM group (Developers), that's why we can temporary disable the option.

Obviously this is not what we want, it's just a temporary work around

@olandese
Copy link

Is there any news about this issue?

@mikebudzynski
Copy link
Contributor

The problem was on the backend side of API Management (cookie wasn't set on redirect) and the fix will be rolled out in the next few days.

@olandese
Copy link

The problem was on the backend side of API Management (cookie wasn't set on redirect) and the fix will be rolled out in the next few days.

I just republished the portal, but the problem is still there. Is the fix rolled out in West-Europe?

@azaslonov
Copy link
Member

Hi @olandese, backend hasn't been deployed to West Europe yet, but it looks like this bug affects only specific services, so, to make sure this is the same case, could you please capture network traces into HAR file and send it to us apimportalfeedback@microsoft.com?

@tomilaakkonen
Copy link

@azaslonov It affects if you force anon users to login. If you disable the option then the problem doesn't occur. We are using West Europe and still having the issue.

@olandese
Copy link

@azaslonov It affects if you force anon users to login. If you disable the option then the problem doesn't occur. We are using West Europe and still having the issue.

Exactly!!!
I described it in a previous comment: #822 (comment)

@LockTar
Copy link

LockTar commented Aug 20, 2020

Why closed? Issues are still there...

@mikebudzynski mikebudzynski reopened this Aug 20, 2020
@azaslonov
Copy link
Member

Looks like it was auto-closed because of PR that associated with it.

@azaslonov
Copy link
Member

azaslonov commented Aug 20, 2020

@LcokTar, after the last hotfix I can't repro this anymore on any of my services. Can you please:

  1. Ensure you have re-published it;
  2. Send service name and network traces to apimportalfeedback@microsoft.com if the issue still there (we need traces from the moment you press sign-in to a one or to loop reloads);

@olandese
Copy link

olandese commented Aug 20, 2020

Hi @olandese, backend hasn't been deployed to West Europe yet, but it looks like this bug affects only specific services, so, to make sure this is the same case, could you please capture network traces into HAR file and send it to us apimportalfeedback@microsoft.com?

Hi @azaslonov ,

I just sent an email with the HAR files and the name of the instances where we are have isues with.
As I told you before, the issue happens when we force anon user to login, see: #822 (comment)

@azaslonov
Copy link
Member

Thanks, we'll look into this.

@LockTar
Copy link

LockTar commented Aug 20, 2020

@LcokTar, after the last hotfix I can't repro this anymore on any of my services. Can you please:

  1. Ensure you have re-published it;
  2. Send service name and network traces to apimportalfeedback@microsoft.com if the issue still there (we need traces from the moment you press sign-in to a one or to loop reloads);

@olandese and I work both on the same 4 instances (with the same issue) sow you have our traces now.

@butsona
Copy link

butsona commented Aug 21, 2020

has this been rolled out globally. The portal is still not refreshing after login. We still have to manually refresh the page to complete login. We are using managed portal in uk south

@tomilaakkonen
Copy link

Any update on this?

@butsona how do you manually refresh? At least for me, f5 after login isn't working. Also clicking on the navigation bar doesn't do any difference.

@azaslonov
Copy link
Member

UPDATE: The issue has been fixed. We're rolling out a hotfix for the backend. It should land everywhere this week. Thank you for patience everyone.

@tomilaakkonen
Copy link

At least in West Europe, we haven't got any updates/hotfixes.

@AnRei123
Copy link

Issue still exists at our location (West Europe). I am still not able to sign in to my published websites. Any news?

@AnRei123
Copy link

AnRei123 commented Sep 1, 2020

2 further Dev instances have been effected now. Seems that someone has by chance pressed the "publish" buttons for these instances. Please urgently come up with a fix!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

@azaslonov
Copy link
Member

@AnRei123, can you send us your service names?
We could try to prioritize them.

@AnRei123
Copy link

AnRei123 commented Sep 1, 2020

Here the names of the API Management service names:

  • apiteamplaydeploytest
  • apiteamplaydev

@azaslonov
Copy link
Member

@AnRei123, can you please give it a try now?

@ErikMogensen
Copy link
Contributor

ErikMogensen commented Sep 2, 2020

It looks like we have the same problem with our instance in West Europe which has:
"PortalVersion":"202009011445",
"CodeVersion":"20200901084820",
"Version":"0.13.1048.0".

Our instance in West Central US also seems to have this problem. It has this version information:
"PortalVersion": "202008260847",
"CodeVersion": "20200812080849",
"Version": "0.13.1056.0"

Edit: Removed info about region updates and put that in a new issue #872

@thgossler
Copy link

The the Chrome browser Application session storage I'm see "Adal" values (see screenshot), and in the Azure Portal I'm seeing a warning that from June 2020 ADAL is not updated and supported anymore. Could it be that there is a problem?

image

image

@AnRei123
Copy link

AnRei123 commented Sep 2, 2020

Retested the two API Management services I mentioned above again:

  • apiteamplaydeploytest: now WORKING again - thank you!

  • apiteamplaydev: still NOT working

@azaslonov
Copy link
Member

azaslonov commented Sep 2, 2020

@AnRei123, the codebase is same for both services, so I suspect there is something cached. Can you please try to reset it:
image

If still doesn't help, please capture Network traces into HAR file: the login attempt itself and several reload; and send it to apimportalfeedback@microsoft.com

@AnRei123
Copy link

AnRei123 commented Sep 3, 2020

For the apiteamplaydev service, the published portal is still not working even after performing all the following steps:

  • Clearing the browser cache
  • Clearing the storage for the unregister service worker options on the application tab card
  • Closing and reopening the browser
  • Republishing again

@azaslonov: I have sent you the HAR file per email.

@AnRei123
Copy link

AnRei123 commented Sep 3, 2020

For the following internal APIM services the sign-in and routing to the homepage via the portal web URL is also not working:

  • apiteamplayuxsource
  • apiteamplayuxplayground
  • apiteamplayuxdefaults
    Could you also fix the issues for these services?

@azaslonov
Copy link
Member

azaslonov commented Sep 3, 2020

Hi everyone, fix deployed, can you please re-publish and try again now?

@thgossler
Copy link

Great, it seems to be working now for us :-)

One thing I had to change in one of our instances was the homepage URL. It was set to the /signin URL as suggested in the tooltip (which is a bit confusing) which resulted in landing on the signin page every time again after sign-in. But this was not the case in other instances where it is now working also.

@ErikMogensen
Copy link
Contributor

Yes @azaslonov, we are now able to log in to both of the portals we couldn't log in to previously. 👍

@AnRei123
Copy link

AnRei123 commented Sep 4, 2020

Thank you for solving this issue! Our portal sign-in works again as expected. Great work!

@azaslonov
Copy link
Member

Thanks everyone for confirmation and your great patience.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
P0 Critical issues.
Projects
None yet
Development

Successfully merging a pull request may close this issue.