It doesn't help when immediately after I make the change (the dark thin bar to the left of the chart) both of the api nodes go down (the big red bit in the chart). :p
So, right now we're running degraded. It should be back with some time, and then the errors should be reduced... > On Sep 28, 2017, at 3:08 PM, Nigel Horne <n...@bandsman.co.uk> wrote: > > Doug, >> Luckily, the response contains the epoch timestamp (1506600623), is my guess. >> >> I've made the health checks and timeouts more aggressive, so Fastly should >> remove the bad node from the load balance rotation hopefully before an >> actual report is sent in. I can't make it perfect, I don't have enough nodes >> to start counting 9s, but I can make it a bit better. >> >> So hopefully this will get a bit better with the new configuration. > Thanks for taking care of this so quickly. > > -Nigel > > -- > Nigel Horne > Conductor: Rockville Brass Band, Washington Metropolitan GSO > @nigelhorne | fb/nigel.horne | bandsman.co.uk | concert-bands.co.uk | > www.nigelhorne.com > > Unless it's for my eyes only, please use "reply all"
