microsoftgraph/aspnet-webhooks-rest-sample

Subscription validation request timed out

Closed this issue · 5 comments

This is the ruby code I'm using to create a subscription in microsoft graph:

        subscribe_endpoint = URI("https://graph.microsoft.com/v1.0/subscriptions")
        http = Net::HTTP.new(subscribe_endpoint.host, subscribe_endpoint.port)
        http.use_ssl = true
        tomorrow = Date.today + 1

        subscribe_request = "{
          \"changeType\": \"created,updated\",
          \"notificationUrl\": \"https://my_url/api/v1/outlook/o365notification\",
          \"resource\": \"/users/#{params[:o365account_id]}/events?$filter=Extensions/any(f:f/id eq 'Microsoft.OutlookServices.OpenTypeExtension.meeteor_event')\",
          \"expirationDateTime\": #{(Time.now + 1.day).to_json},
          \"clientState\": \"SecretClientState\"
        }"

        subscribe_response = http.post(
            "/v1.0/subscriptions",
            subscribe_request,
            'Authorization' => "Bearer #{params[:o365account_access_token]}",
            'Content-Type' => 'application/json'
        )

it worked great, but suddenly today I'm getting a Subscription validation request timed out response. Altough when I'm doing a POST to https://my_url/api/v1/outlook/o365notification it's just working fine.

Any ideas why it stopped working when nothing changed on my side?

Could it be that, because my_url points on my dev environment and is often offline, the domain got banned?

Hi @alekkowalczyk . Since setting up my notification endpoint, I've only gotten that error when my service is offline (which also happens a lot). I just retested and I'm able to create a subscription.

Any chance your response time has slowed down?

I forwarded your question along to see if we can troubleshoot on this end.

Hi @DianeD thanks for the quick response.
I don't think the response time slowed down, when I do an POST call to the endpoint from Fiddler, the response is quick, it's a dev environment so it can take up to a half second but nowhere long enough to do an timeout if it's not set to extremely low values.

Issue was resolved on StackOverflow:

"If anybody is interested: after our offline follow up it turned out that it's a connection issue, after my endpoint was deployed on a cloud service it worked. For endpoints pointing to dev machines ngrok is suggested."

Thanks for reporting, @alekkowalczyk !