Facebook
TwitterHello
We're running Plesk on a Virtual Server and what we're finding is quite strange, usually every night roughly between the hours of Midnight (00:00) and 8am (08:00) for periods of time anywhere from a few minutes up to 4 hours our the Apache Service will timeout for a lot of or all HTTP Requests.
For example during the normal working day everything is fine, no issues (which is ironically when we would expect to get the most traffic), but last night through the night there were periods of a few minutes where the server stopped responding, then a long period of 2 hours 44 minutes where the server stopped responding to requests. This pattern continues each night but the times and durations of the timeouts vary and don't seem to be predictable.
What makes this more strange is that the level at which the timeouts occur can vary between vhosts. For example the described timeout pattern above happens for one vhost but another vhost we are monitoring has a different pattern of timeouts, for this other vhost the periods at which the server does time outs don't last as long, and when they do happen they seem to consist of intermittent timeouts where the server will respond fine for a few minutes then stop responding for a few minutes and continue that pattern for a while then everything will be fine again.
Update 14 May: So far we have managed to narrow it down to the Apache Server/Service, we have done this by monitoring different vhosts and pinging the exact same page on each vhost, with the difference being one vhost is just running with just Nginx (PHP FPM served by Nginx with Proxy Mode off), while others are running with Apache (PHP FPM served by Apache with Proxy Mode on).
I have attached a screenshot of a sample of the uptime over the past 24 hours to illustrate what I'm talking about, the green bars are where the server responds with no issues, the red bars are where the server is timing out. The first graph is the second vhost I mentioned and the second graph is the first vhost I mentioned.
Below this message I have included an extract from the Plesk Log Viewer of the first vhost I mentioned for when the timeouts do occur.
Any other time when the server isn't timing out the logs look normal.
The resource usage of the Virtual Server is always low, even during these times and we've even tried moving it to a different physical host that is under utilised, but doing that made no difference.
Any advise or suggestions would be much appreciated, the fact that there seems to be very little consistency to this and that the server resource usage is never high enough to cause problems is what is throwing me off mostly.
Thanks
Aaron
-- Log Extract --
We're running Plesk on a Virtual Server and what we're finding is quite strange, usually every night roughly between the hours of Midnight (00:00) and 8am (08:00) for periods of time anywhere from a few minutes up to 4 hours our the Apache Service will timeout for a lot of or all HTTP Requests.
For example during the normal working day everything is fine, no issues (which is ironically when we would expect to get the most traffic), but last night through the night there were periods of a few minutes where the server stopped responding, then a long period of 2 hours 44 minutes where the server stopped responding to requests. This pattern continues each night but the times and durations of the timeouts vary and don't seem to be predictable.
What makes this more strange is that the level at which the timeouts occur can vary between vhosts. For example the described timeout pattern above happens for one vhost but another vhost we are monitoring has a different pattern of timeouts, for this other vhost the periods at which the server does time outs don't last as long, and when they do happen they seem to consist of intermittent timeouts where the server will respond fine for a few minutes then stop responding for a few minutes and continue that pattern for a while then everything will be fine again.
Update 14 May: So far we have managed to narrow it down to the Apache Server/Service, we have done this by monitoring different vhosts and pinging the exact same page on each vhost, with the difference being one vhost is just running with just Nginx (PHP FPM served by Nginx with Proxy Mode off), while others are running with Apache (PHP FPM served by Apache with Proxy Mode on).
I have attached a screenshot of a sample of the uptime over the past 24 hours to illustrate what I'm talking about, the green bars are where the server responds with no issues, the red bars are where the server is timing out. The first graph is the second vhost I mentioned and the second graph is the first vhost I mentioned.
Below this message I have included an extract from the Plesk Log Viewer of the first vhost I mentioned for when the timeouts do occur.
Any other time when the server isn't timing out the logs look normal.
The resource usage of the Virtual Server is always low, even during these times and we've even tried moving it to a different physical host that is under utilised, but doing that made no difference.
Any advise or suggestions would be much appreciated, the fact that there seems to be very little consistency to this and that the server resource usage is never high enough to cause problems is what is throwing me off mostly.
Thanks
Aaron
-- Log Extract --
Code:
12/04/2018 07:21 Access 301 GET HTTP/1.0 178 nginx SSL/TLS access
12/04/2018 07:21 Error 1425#0: *20680 upstream timed out (110: Connection timed out) while reading response header from upstream nginx error
12/04/2018 07:21 Error (70007)The timeout specified has expired: AH01075: Error dispatching request to : (polling) Apache error
12/04/2018 07:22 Error (70007)The timeout specified has expired: AH01075: Error dispatching request to : (polling) Apache error
12/04/2018 07:22 Error (70007)The timeout specified has expired: AH01075: Error dispatching request to : (polling), referer: Apache error
12/04/2018 07:22 Access 200 GET HTTP/1.0 8.48 K Apache SSL/TLS access
12/04/2018 07:22 Error (70007)The timeout specified has expired: AH01075: Error dispatching request to : (polling) Apache error
12/04/2018 07:22 Access 301 GET [URL] HTTP/1.1 178 nginx SSL/TLS access
12/04/2018 07:22 Error (70007)The timeout specified has expired: AH01075: Error dispatching request to : (polling), referer: [URL] Apache error
12/04/2018 07:22 Error (70007)The timeout specified has expired: AH01075: Error dispatching request to : (polling), referer: [URL] Apache error
12/04/2018 07:22 Access 200 GET [URL] HTTP/1.0 8.37 K Apache SSL/TLS access
12/04/2018 07:22 Access 301 GET [URL] HTTP/1.0 178 nginx SSL/TLS access
12/04/2018 07:22 Error 1425#0: *20688 upstream timed out (110: Connection timed out) while reading response header from upstream nginx error
12/04/2018 07:22 Access 302 GET [URL] HTTP/1.0 792 Apache SSL/TLS access
12/04/2018 07:22 Access 302 GET [URL] HTTP/1.0 792 Apache SSL/TLS access
12/04/2018 07:22 Access 200 GET [URL] HTTP/1.0 2.10 K Apache SSL/TLS access
12/04/2018 07:22 Access 200 GET [URL] HTTP/1.0 2.10 K Apache SSL/TLS access
12/04/2018 07:22 Error (70007)The timeout specified has expired: AH01075: Error dispatching request to : (polling) Apache errorAttachments
Last edited:
