Following the failure/retirement of one of our oldest / legacy voip nodes last night it has become apparent that a number of customers have hard coded either the IP address of that node or the DNS for voip-3 instead of the DNS sip.fidotalk.co.uk for their SIP server
Further to this we are aware that some customer’s equipment has not refreshed the DNS since we updated the DNS to remove voip-3 from the DNS cluster at 22:00 last night. If you are experiencing registration problems this morning then it is quite possible that these are DNS cache related and we would recommend a forced flush of your DNS cache.
We tend to use a 5 minute TTL for DNS entries on front end services which should mean that your equipment refreshes regularly and spots any changes however in this instance it appears that not all equipment is honouring TTL entries in the DNS.
We are working to implement a hardware load balancer in front of the DNS cluster instead of (as well as) using SRV records and DNS entries in order to try to mitigate any similar issues going forwards.
If customers are still experiencing problems with calls we would ask you raise a support ticket providing as much information as possible and we will investigate the individual reports.