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. There's been 3-4 new testers recently as well, but our problem is not volume of report submissions (it's still the performance of other things affecting the system as a whole). > On Sep 28, 2017, at 2:31 PM, Nigel Horne <n...@bandsman.co.uk> wrote: > > When I was submitting a report. > > Metabase submission doesn't timestamp, it just says things like "Submit > '0OZncYKk5xGxvTcp8awfHg==' (4.2263879776001s) success" > > -Nigel >> Is this when you're submitting a report or when you're doing something else? >> Was that message at approximately Thu Sep 28 12:10:23 2017 GMT? >> >> >>> On Sep 28, 2017, at 7:14 AM, Nigel Horne <n...@bandsman.co.uk> wrote: >>> >>> I've been seeing a number of messages such as this: >>> >>> <?xml version="1.0" encoding="utf-8"?> >>> <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" >>> "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd";> >>> <html> >>> <head> >>> <title>503 backend read error</title> >>> </head> >>> <body> >>> <h1>Error 503 backend read error</h1> >>> <p>backend read error</p> >>> <h3>Guru Mediation:</h3> >>> <p>Details: cache-dca17747-DCA 1506600623 2675462800</p> >>> <hr> >>> <p>Varnish cache server</p> >>> </body> >>> </html> >>> >>> >>> -Nigel > > Wh > > -- > 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"
