We never dropped the service since we were asked to install it because
we had too many CPUs few years ago.
I've even spent quite some time keeping it going when it was crashing
one day yes and one day no 1.5 years ago.
It felt silly to drop it now that it is stable and doesn't require much
work maintaining it. :)
On 16/11/2012 12:19, John Gordon wrote:
> I thought the UK recommendation was RAL+Imperial. I think someone monitored it and we all had >1. EGI actually monitor the combined availability of RAL.OR.Imperial TL BDIIs. As you might expect it is always 100%.
> The only problem was the nagios tests which only used RAL so they failed when a real job wouldn't have.
> -----Original Message-----
> From: Testbed Support for GridPP member institutes [mailto:[log in to unmask]] On Behalf Of Alessandra Forti
> Sent: 16 November 2012 12:00
> To: [log in to unmask]
> Subject: Re: GDB 14th Nov 2012
>> In fact EGI (Tiziana) advocates having multiple servers behind a DNS alias - it seems the UK is unusual in not doing it like that. I'm not sure if there's much practical difference - anyway it seems that most top BDIIs, including ours, are highly reliable now. But obviously something like the recent powercut at RAL implies you must have some kind of cross-site faiilover - did that cause problems at any UK sites?
> most other countries probably don't have distributed failover but
> multiple top BDIIs at T1. Manchester didn't have any problem during the
> power cut. I'm actually glad we didn't drop our service in the move to
> EMI-2 for once it was useful.
Facts aren't facts if they come from the wrong people. (Paul Krugman)