dring takes 100% CPU when calling a non-available ID
Issue generated from Tuleap's migration script. Originally submitted by: Baptiste Jonglez (baptiste)
The "dring" process often takes 100% CPU when calling a ring ID that is not currently available. How to reproduce (it does not always reproduce, unfortunately): - using the GNOME client, call the ID of a contact that is not currently connected (or a random ID) - wait for the DHT resolution to fail (message goes from "Searching for" to "Failed") - observe then that the "dring" process takes 100% CPU in htop - when pressing "Hang-up", the high CPU usage stops Sometimes, the high CPU usage occurs while DHT resolution is running ("Searching for" in the GNOME client). Sometimes the high CPU usage remains even after the call has been hanged up. It seems that trying to call multiple ID at the same (or even the same ID multiple times in parallel) increases the chance to reproduce. I am using opendht 0.6.3, ring-daemon 082d9dd14a5b334fd6ddb2429ef1d7bbfdcca15c, libringclient d8710c09e1c982a60f6063c2cd3742407a6415bc, and ring-gnome 541a40ab.