We have an assistant built using the bot framework (V3 SDK). It is connected to the Facebook channel.
Within the Facebook page setup, we have a group of testers configured.
One of our testers does not receive a response from the bot. Other testers do. We have removed the user from the test group, and have readded him (same as our other testers who are working fine), and still no responses.
The messages from the user show in the page's inbox.
Our logging shows that it never gets to the bot framework though.
Note, we have 10+ testers, and this is the only user that has this issue. All others, are able to send and receive messages as expected.
When I connect using ngrok, the messages are routed properly, and the conversation occurs as expected. I then disconnect ngrok, and the user is no longer able to have a conversation again.
The user was able to test for the previous 8 months, and this issue just started about two weeks ago.
I am at a loss for what it could be.
Related
I have a Slack bot that is working fine and interacting with users. I'm using Bot Framework composer and the Slack Adapter.
In the Slack API portal I'm trying to change the Events Request Url the app uses to send Slack Events to my bot.
When I do that, slack sends a challenge request to my bot. The bot first tries to verify that the request is really coming from Slack following: https://api.slack.com/authentication/verifying-requests-from-slack#a_recipe_for_security
The problem is that this is failing and I can't understand why.
I see that Slack is sending all the right content, and that the ClientSigningSecret is being read, otherwise the other calls to the bot wouldn't work.
I know it's a bit far fetched to ask this since it seems to be a problem on my side. But since the bot is validating the requests just fine when users talk to the bot, and the code is from the Slack Adapter which is open source and there's nothing else I can thing of..... maybe someone struggled with the same problem.
I created a support ticket to Slack and they came back pretty quickly.
Pre publish state
Before publishing a Slack app the only configs that exist are the ones you see in the App configuration page. Those are what you use to test your app, this includes the secrets to authenticate the incoming messages from Slack into your backend.
After you publish your Slack App for the first time
Once your app is published, the production version that your users use will see the original settings, including the secrets and these are the ones your backend will get.
The settings you see in the configuration page are like development mode and they won't be persisted into the published app until you request Slack to approve your changes. That's sounds great and is what one would expect, but what you don't see and have no way of imagining is happening is that there are some development time secrets that are different from the ones you see on the settings screen.
When you change the endpoint url to be sent to your backend so that it can return the challenge and Slack would accept the new url, the message payload goes with this development secret and not the one you configured your backend with. Thus your backend will reject the call since it thinks it's not coming from Slack.
Proposed solution from Slack
Don't validate the signature of the incoming request for this type of call in an already published app. I don't like it but there was no other workaround unless Slack changes this. So what I did was:
Remove that check only for this request from the backend and publish to production.
Make the url change in Slack.
Revert the change from the backend.
:(
I'm trying to build a personal Teams bot which has to send proactive messages regularly. I made it work by storing the conversation when the user installed the app, just like the examples here. Now it works great, but I realized that I keep messaging users that have uninstalled my application. I can't seem to find any event that is triggered when a user uninstalls a personal scoped bot (app).
I tried conversation events, but nothing seems to be triggered for personal scoped bots at least.
I'm using the NodeJs botbuilder SDK.
Open to any suggestions, thanks.
To my knowledge, there is no such event unfortunately. I think you'll just receive a 403 at the moment without much info. This is due to change though - Microsoft just made an announcement about this yesterday in fact. See https://developer.microsoft.com/en-us/office/blogs/best-practices-and-updates-for-bot-lifecycle-events-in-microsoft-teams/ in the paragraph "Changes to post uninstall behavior for bots in personal scope":
Moving forward we’ll also align the post uninstall behavior for bots in the personal scope with the teams and groupChat scopes and you will not be able to send or receive messages after an app has been uninstalled. Your bot will receive a 403 response code to new messages posted by your bot. The 403 response code will have fields telling you the reason behind the it – which is either the app was uninstalled or the bot was blocked. We expect these changes to roll out in the coming few months.
I have a Microsoft bot app that can be installed in Teams on a team.
For now, I save the tenant ID + team id in our DB when the app gets installed, via a received activity on our end point.
What happens if my prod is not up at the moment of the installation? (for whatever reason, could be an update or just a technical downtime...)
Is it lost forever? Is it retried?
Same question with a 500 error?
This activity is an example, it could be any other activity. Ex: a user is added/removed, a message has been sent etc...
For OnMembersAddedAsync; unfortunately there is no retry from the Teams side, it fires and forgets.
The information in that Activity will be included in any subsequent send from Teams - so the bot can recover, if the user interacts with the bot again. Messages would just be sent when the bot recovers.
You can find a bit more info on Teams and events here:
How do I send a 1:1 welcome message?
We are using a web application to send out mail. Email is sent and received as expected on a server level and in GMail. However, when messages are pushed to our client's Office365 (Outlook), instead of getting one copy they are getting duplicates.
This is how the app works. You can send emails to a number of people (1,5,20, etc). It's all the same message, subject, etc. It's the same email.
The app sends you a copy. Just one. That's how it is supposed to work. Within their Office365 (Outlook) email account, they get their one copy, and then another copy for each recipient (that's not supposed to happen). So, if you email 5 people, instead of getting one email back, they get 6.
The thing is, this program has run flawlessly for nearly a year. Out of the blue, this glitch started happening.
We have had no code changes, no vendor updates, or anything of the sort on the web application side of things that would trigger since before the time this glitch started happening. When we tested locally, and on the live server, with GMail credentials, the app continues to work as it should.
We are sending out mail via SMTP.
The security protocol is "tls".
https://laravel.com/docs/5.5/mail that describes the core Laravel mail function we are using to send out mail.
The client contacted Microsoft support -- who basically said "I don't know. Not my problem".
I don't know where else to turn, or what other steps to take to debug.
I've built a Bot using C#. I tested it using the emulator, web chat, direct line and Skype, and it's working as expected.
I followed the steps to register the bot with skype for business, I waited more than 24 hours to see what's happen.
So far, I can see the Bot as a contact, however, when I try to send a message, the error
"Error happened in Skype for business when reaching bot service. We
saved this conversation. You'll see it soon in the Conversations tab
in Skype for Business and in the Conversation History folder in
Outlook."
Looking for some logs or something, I collected some info, that I'm sharing here.
Analytics from BOT that proves that it's reaching it:
The Log for SFB channel said: "There was an error sending this message to your bot: HTTP status code NotFound"
It's look that the Bot endpoint is not available. When I registered it in SFB, I've been using NGROK (to run agains my machine), but later I've change it to a azure site.
"Old" endpoint is used yet today:
So, all I can guess is that Bot End Point was "registered" at the moment that I made the registration of my Bot in SFB, and now it's not possible to change it.
Does it make sense to anyone of you?
My problem seems to be kind of similar to Bot Framework - An error while sending a message from Skype for Business
Adrián
Ok, I'm sure the problem with the bot was the endpoint registration. I found a powershell script that update the endpoint.
I Opened Windows PowerShell as Administrator and run the ff scripts:
Import-PSSession (New-CsOnlineSession -Credential (Get-Credential))
Set-CsOnlineApplicationEndpoint -Uri sip:username#yourdomain.com
Obviously, I changed username#yourdomain.com with my own.
Then, after waiting 8 hours (or so), I was able to communicate with the bot, using SFB as channel.
Bot running as expected
I hope this can be useful for others.