1

For the past week or so our CentOS server (that runs Apache and NGIX as a proxy) has been acting odd as its bandwidth output has been consistently dropping from around 10/11pm until 4am to around 1M (neither dipping or raising past that). Here's a graph to illustrate what happens.

https://i.stack.imgur.com/G4JEe.png

Our server runs a handful of sites, one of which is my own that runs Wordpress. When the bandwidth output drops all sites on the server run slow which has lead us to assume something is hogging the bandwidth.

We've checked all of the crons running at the times the bandwidth output drops but nothing is out of the ordinary. Also if you require any more information about the server, such as its configuration, I'll try and provide what I can.

Any help or advice would be greatly appreciated, also if there's a better place for me to post this let me know :)

Here is httpd usage

https://i.stack.imgur.com/yMCUY.png

And Nginx usage

https://i.stack.imgur.com/YLxdi.png

This happens between 22:45 till 04:00 on these graphs

Oliver Whysall
  • 195
  • 1
  • 6
OWhysall
  • 11
  • 1
  • Where is this server and what kind of network and internet connection does it have? – Grant Oct 03 '14 at 15:00
  • The server is currently in an office with a UK Virgin Media Business line with dedicated static IP coming into it, the internet is just coming straight out of the modem straight into the server – OWhysall Oct 03 '14 at 15:03
  • 2
    Have you talked to virgin? Perhaps they have been doing network maintenance around those times? – Grant Oct 03 '14 at 15:07
  • I have spoken with Virgin and they can confirm there is no work being carried out or throttling on the account – OWhysall Oct 03 '14 at 15:18
  • Isn't it possible that the server is just not busy/under load during that period of time? – joeqwerty Oct 03 '14 at 15:46
  • It seems to be the same time every night like clockwork – OWhysall Oct 03 '14 at 15:58
  • 2
    Can you correlate this to a graph of actual requests? Maybe the requests are lower during this time? – mfinni Oct 03 '14 at 17:44
  • I have added the relevant graphs to the post that correspond to the time of bandwidth outage – Oliver Whysall Oct 04 '14 at 18:40
  • Have you had a look at your other machines graphs too? Anything like outstanding CPU usage? Nothing seems wrong with the graphs you provided and nginx connections fluctuation looks pretty normal to me. If you cannot see anything wrong with your machine then it is network-related. Ping your provider back again. – Bernard Rosset Nov 06 '14 at 12:47

0 Answers0