[_] Performance testing, please help!
stephen.doyle at mobile-life.com
Mon Feb 4 18:45:47 GMT 2013
Have you tried loading the site from various locations using apache benchmark - that could help determine if it's a localised client / ISP problem or a server issue? VPS hosts can have the performance drop off alarmingly when under moderate concurrent load and apache benchmark can help discover the relationship between concurrent requests, response time and connection problems. I assume you're hosting with a VPS provider and your host might be swapped out at times. I would try increments of 5 concurrent requests running 10 cycles per test and see what happens between 5 and 30 concurrent requests. Steve On 04/02/2013 17:58, "Tim Wintle" <timwintle at gmail.com> wrote: >On Mon, 2013-02-04 at 17:39 +0000, Adam Nicholls wrote: > ><snip> > >> To clarify: when I say images/css not loading I mean there's a spinning >> wheel next to those files in Firebug. I've done a 'netstat -an' on the >> server and can see the browser has made a connection, although I've >>noticed >> a few "TIME_WAIT" statuses, not sure what this indicates though or if >> they're related to the files! > >That just means the socket is effectively closed, but it's waiting just >in case it needs to re-submit the acknowledgement that the socket will >close (i.e. you can ignore them unless you're maxing out your open >sockets. > > >It is possible that there's a networking issue - if you run wireshark on >the client's computer you would be able to track down exactly what's >going on here in more detail, but you'll need a good knowledge of Layer >3 networking (i.e. TCP/UDP). > >You may be maxing out allowed sockets on either the client or server, >but you'd need to look at full network logs to find that out. > >> Anyone else experienced this kind of behaviour before from a Server or >>ISP? > >Yup - from servers, NICs, switches, datacenters, trans-atlantic >cables... > >I'm afraid networking is a dark art and I'm not an expert. > >Tim > > >-- >underscore_ list info/archive -> >http://www.under-score.org.uk/mailman/listinfo/underscore