action=tally&value=all causes slower httpd response

[fabbietto]

Hi,

i have a server with almost 150 domains so is not overused (4gb ram, quadcore, load 1-2 at all), i notice that at midnight when action=tally&value=all is adding to task.queue the httpd response is very slowly and is possible to wait 10-15 seconds to see domain's homepage, while during the day every site is open in 1-2 seconds.

Anyone has the same problem?

Thanks

 

[zEitEr]

Hello,

I/O? IOwait? HDD dying?

 

[fabbietto]

I tried to stress the server but i'm not able to reproduce this slowly response like during tally&value, it's a virtual server , i also try to move it from his node to exclude virtualization problem but nothing,

 

[zEitEr]

What if you run a backup of all users of yours during a daylight? With nightly tally what do you have there? Do you run backuping? Or not?

 

[fabbietto]

no other process is running during tally, no backup, no cron, no massmail..
for now i simply modify cron from 00:10 to 03:00, in this way is more difficult to see the problem, in my country obviously

 

[zEitEr]

Is Awstats enabled? Webalizer? With nightly tally directadmin update their stats data.

 

[gate2vn]

Try command "ps aux" to find out if other command running in a long time. How about the server load at that time too? From top command?

With that VPS, how many percent of CPU can you use?

I tried to stress the server but i'm not able to reproduce this slowly response like during tally&value, it's a virtual server , i also try to move it from his node to exclude virtualization problem but nothing,

 

[fabbietto]

At the moment i modify cron to 2 AM, in a few months i'll make a firmware update on my storage, i'll reply if this will solve the problem, thanks all