High CPU usage

DoM

Well-Known Member
#1
Hi,
for days litespeed works very well. Load always less than 2.

From this afternoon load reach about 60 and ALL cpus, (8 cores) were taken from:

nobody 28718 20.0 2.0 186996 170028 ? R 18:20 0:00 | \_ litespeed (lshttpd)
nobody 11830 95.7 1.7 161684 145860 ? R< 17:17 28:29 \_ litespeed (lshttpd)
nobody 11831 95.6 1.7 161684 145860 ? S< 17:17 28:26 \_ litespeed (lshttpd)
nobody 11832 95.8 1.7 161684 145860 ? R< 17:17 28:35 \_ litespeed (lshttpd)
nobody 11833 96.0 1.7 161684 145860 ? R< 17:17 28:43 \_ litespeed (lshttpd)
nobody 11834 95.7 1.7 161684 145860 ? R< 17:17 28:32 \_ litespeed (lshttpd)

So i would like to understand when it happens and how to fix it.

I checked on stats viewer on litespeed but i do not see any abonormal situation..

Stats says that _SERVER still receive about 50-100 req/s but on access_log i cannot see any particular request to server so i would like also to understand if all those req/s were registered somewhere else for server or always on /usr/local/apache/logs/access_log


Waiting for your reply


Regards
 

webizen

Well-Known Member
#4
If /usr/local/apache/logs/access_log recorded requests in the past but could not when system load was high, it is possible that some processes got stuck and others had to wait. Hence no log was recorded. You may run a quick test like issue 'curl localhost' (whatever the webserver listens to) and see if anything comes back. If it does, check anything gets logged.
 

DoM

Well-Known Member
#5
Hi,
did it but nothing gets logged.

I also checked that on litespeed > web console > general

there were these 2 directives:

Log Control: Own Log File

File Name: $VH_ROOT/logs/access.log

But nothing gets logged.


Waiting for your reply


Regards
 

webizen

Well-Known Member
#7
Since LSWS loads apache configurations, requests get logged in /usr/local/apache/logs/access_log. We need to see what happens when cpu load gets high. Is there a pattern on what time during a day the load becomes high?
 
Top