Adam Bower adam@thebowery.co.uk wrote:
On Tue, Sep 07, 2010 at 12:58:23AM +0100, MJ Ray wrote:
we'd've liked it to be better this year, but hardware failures occur at random. Following the above-mentioned problem, we've decided to
To be honest, many hardware problems manifest themselves (especially disk problems) days or weeks before they take a machine down. Try monitoring your disks more carefully and get better servers with hardware health monitoring and use it.
We have some hardware monitoring (and we're always considering better servers) and we use it. When we see a fault developing, the hardware gets replaced. This wasn't one of them. Even when the dang thing was failing, it wasn't clear which disk was faulty. That's why it was so disruptive.
Does anyone have suggestions for other actions that would push that uptime higher while keeping the cost reasonable?
Buy more servers and have some kind of cluster.
Need more clients for the cost of that to stay reasonable. In preparation for that, the next server will have some configuration changes to make moving hosting clients around between servers quicker and easier, to help move them onto a cluster.
That's something which is still remarkably without a standard format, as far as I know. Or have I missed a development and there's a standard hosting settings file format now?
Regards,