|Concerned about a server load of 9.50|
| 10:47 pm on Jan 18, 2006 (gmt 0)|
I have a resellersaccount on a shared server:
Processor #1 Vendor: GenuineIntel
Processor #1 Name: Intel(R) Xeon(TM) CPU 2.80GHz
Processor #1 speed: 2793.503 MHz
Processor #1 cache size: 1024 KB
Processor #2 Vendor: GenuineIntel
Processor #2 Name: Intel(R) Xeon(TM) CPU 2.80GHz
Processor #2 speed: 2793.503 MHz
Processor #2 cache size: 1024 KB
Processor #3 Vendor: GenuineIntel
Processor #3 Name: Intel(R) Xeon(TM) CPU 2.80GHz
Processor #3 speed: 2793.503 MHz
Processor #3 cache size: 1024 KB
Processor #4 Vendor: GenuineIntel
Processor #4 Name: Intel(R) Xeon(TM) CPU 2.80GHz
Processor #4 speed: 2793.503 MHz
Processor #4 cache size: 1024 KB
Last night I saw server loads between 8.50 and 9.50. This lasted for a few hours.
I reported this to the webhost, since I thought the sites hosted on there were slow as well, and I was worried about the high figures, since normally it's not even above 1.00.
The host reported back to me and said server loads of 8-9 are no problem, since it is a 4 cpu machine. What do you think?
| 6:41 pm on Jan 19, 2006 (gmt 0)|
The load average under unix/linux (as I think what you mean by the server load) is the average number of processes in the running queue. On a 4 processor machine four processes can run simultaneously, so a load of 9.5 would indicate that every process would run at about 40% of the speed compared when it was running dedicated on a 1 processor system.
So in my opinion a load average of 9.5 is very acceptable for a four processor configuration. I have seen internet servers running with load averages of 600 to 700, and yes those ARE slow, but they were still running and performing their tasks: mainly serving FTP requests for large downloads.
| 4:50 am on Feb 5, 2006 (gmt 0)|
The rule of thumb is 3 times the number of processors... So, if it is consistently above 12 for your server, you may want to find a new host.
| 4:40 am on Feb 6, 2006 (gmt 0)|
When I check my server uptime before the crash, I saw these numbers.
[admin@srv1 admin]$ uptime
20:32:03 up 159 days, 14:17, 1 user, load average: 4.xx, 4.xx, 4.xx
Sorry I couldn't remember the numbers after the decimal... but they are all around 4. Sometimes 5-6.
How come my server can crash between 4-6 when you guys can still survive at 12, or even higher.
My server config:
Pentium 4 1.7Ghz 400Mhz FSB
My site doesn't get a lot of traffic... around 30,000 to 35,000 page views a day. All pages are dynamically generated. I also request 2-3 XML query to another server for every page load.
| 5:30 am on Feb 6, 2006 (gmt 0)|
A server crashing often means it ran out of memory. Any idea what memory usage was prior to the crash?
Also, load average due to CPU usage is one thing, if it is due to iowait, things can get bad very quickly. A load of 12 on a 4 CPU machine purely due to CPU usage will slow things down but not be too bad. Same load on same machine caused mainly by iowait is a different story.
| 7:21 am on Feb 6, 2006 (gmt 0)|
Well, I got these now when retrieving the memory info using 'free'
total used free shared buffers cached
Mem: 1025376 965072 60304 0 131240 404220
-/+ buffers/cache: 429612 595764
Swap: 2048276 56576 1991700
But that's nowhere near a crash... what do you think?