Wednesday 23 July 2008

Nameserver (DNS) upgrades

As part of our rolling upgrade cycle, new nameservers are being brought online and older ones retired. For all domains that we manage these changes will happen automatically, and there is no expected impact on any services.

Monday 21 July 2008

Update From Friday 18/07/2008

Gladserv suffered a prolonged outage on Friday afternoon affecting our Manchester servers. The root cause was a hardware failure on a server that had been running for over 470 days. Some web, email and database services were affected for up to four hours. Normally under these circumstances we would failover to another datacentre in London or Edinburgh.

However, attempts to move services to the Edinburgh datacentre were hampered by by high load on those servers. New servers had been installed in London and Edinburgh but were not yet in service. Earlier in the afternoon we had installed a new server in a separate Edinburgh facility to take some of the load, but this server had literally just been plugged in when the Manchester outage happened and so was not ready.

To recover from the outage, an engineer was despatched to the Manchester datacentre to replace the faulty hardware while we worked at preparing the new server in Edinburgh. The Manchester systems were brought back online just as the Edinburgh server came into service.

By early Friday evening all systems were online, including new servers in London and Edinburgh, and an upgraded Manchester server. We continued to work through the weekend to ensure we can recover much more quickly from a similar outage in future. On Sunday evening we performed a firedrill for the most critical systems that failed on Friday and are pleased to say that all went to plan.

Thank you to our affected customers for your patience throughout.



Brett Sheffield
Managing Director

Friday 18 July 2008

Manchester Data Centre Connection Problems (SOLVED)

Affecting some email and web services.

London and Edinburgh datacentres are unaffected.

We are currently investigating the cause and hope to have this resolved shortly.

It appears to be a hardware issue, we will update one this has been solved.

More to come soon.