[OSM-talk] [Announce] Server maintenance – 26th October

[Tom Hughes]

On 20/10/14 20:06, Andy Mabbett wrote:
> On the announce, list, on 20 October 2014 18:38, Harry Wood
> <mail at harrywood.co.uk> wrote:
>
>> Thank you, in advance, for your patience and understanding during the maintenance window.
>
> Thank you and the team for keeping things running.
>
> I in no way mean this as a criticism, but am asking out of interest:
>
> What would it take to make our system so robust that such outages were
> not required? Is it a case of more money, or more developers, or both?
> Or is it entirely infeasible?

In theory we can do the master/slave swap without any downtime, and I 
try and do it that way, but to be absolutely safe I will probably do it 
with the site in offline mode for a few minutes. That way I can also 
take the opportunity to reboot the servers to update the kernels.

The announced window is really an "at risk" and I don't expect any 
outage to be more than a few minutes if all goes to plan.

GPX trace uploads may be down for a bit longer, as something else has 
come up today which means they need to be taken offline while a 
filesystem is resized and I may try and do that in the same window.

Tom

-- 
Tom Hughes (tom at compton.nu)
http://compton.nu/