A few days ago I encountered a strange problem in a customer’s testing environment. The environment consisted of two enterprise pools with 3 Front Ends in each pool running Skype for Business Cu3. Signing in from a mobile device had stopped working for some reason.
Inspecting the logs I could see that authentication worked as expected and the clients all received their webtickets properly but clients still wasn’t able to sign in. Looking at the client logs i found the following.
INFO APPLICATION CAlertReporter.cpp:64 Alert received! Category 2, Type 300, level 0, error E_UcwaUnavailable (E2-3-48), context ‘handleUcwaAppSessionRequestError’, hasAction=false
So obviously the client didn’t get the expected response from UCWA. Inspecting the Mobile Gateway (reverse proxy) logs we could indeed see that when our client tried to access /ucwa/v1/applications it received a HTTP 503 Service Unavailable response from the Front End. Accessing the URL directly on the Front End also gave us the 503 response. When accessing the URL directly on any of the other Front Ends in the pool we received a 401 Unauthorized response which is the expected response.
We issued a restart of the web server on the affected Front End from the IIS manager and tried again. Unfortunately this didn’t solve the problem, the clients still received the 503’s. I’ve seen issued before with a restart issued from the IIS manager. So we tried again using an iisreset issued from a command prompt and… Voilá, the client was able to sign in and direct access to the URL now gave a proper 401 response.
Conclusions: resets from issued from the IIS manager does not really restart and flush the web server. While I still don’t know exactly what caused the 503’s I believe it might be related to system resources, as this was a testing environment the machines had limited resources.