Load average


#1

hello guys…
i recently move my sites to dreamhost… i take the machine nilla and especially in this month (july) occur a lot connections problems with the server and because this i ask to dreamhost team move my account to another machine…
now i’m in limbo-spunky3 and there’s no connectins problems… by the way my sites are very slowly to process…
when i check the load average i’ve got

daniel_com@limbo-spunky3:~$ uptime
12:38:30 up 15:54, 4 users, load average: 209.44, 221.77, 225.20

it’s not normal… nilla has more users than limbo-spunky3 and has load about 5…

anyone has the same problem.!? or know why it’s happening?


#2

I’d suggest letting DH know through the panel, as that is waaaay too high. Someone is probably running a terrible script or getting way too much traffic.

Have you tried “top -b -n 1” to see a list of the processes? If you see a particular user or script that is hogging a lot of resources, you could add that to your contact with DH to give them a heads up. Good luck!

Check out Gordaen’s Knowledge, the blog, and the MR2 page.


#3

Since each user pretty much runs in their own sandbox, ‘top’ won’t show you what other users are doing.

-Scott


#4

yeah…
top shows only my proccess… not another users process


#5

I reported a system high load problem of bazooka too with unlimited “top”.

After getting a email from DH support, I was put into sandbox. I don’t know if they want to hide something or something.


#6

Peeking at other users’ processes may include passwords embedded into a running command.

-Scott


#7

Ah, good point, I spaced on that one!

I think the load averages shown with uptime are average number of processes running or waiting to run for 1min, 5min, 15min. Is that correct? If so, the numbers are useful for general purposes. But 200+ is insanely insane!

Most people sysadmins I’ve talked to say it should be around the number of processors the system has (same say double, some say # + 1). So even with the higher figures 2 cores on 2 processors x2 is still only 8.

Hopefully DH takes care of whatever is causing the crazy load.

Check out Gordaen’s Knowledge, the blog, and the MR2 page.


#8

[quote]An idle computer has a load number of 0. Each process that is using CPU, waiting for CPU or is in uninterruptible sleep (usually waiting for disk activity) adds 1 to the load number.

The load average is calculated as the exponentially damped moving average of the load number. The three values of load average refer to the past one, five, and fifteen minutes of system operation.[/quote]
I’ve always interpreted load average to mean the amount of processes that could be running but aren’t because they’re waiting on a busy CPU.

I’ve read that acceptable load averages should be below the number of CPUs * 3. So on a 4 processor machine the load average should remain below 12, otherwise you could consider it overworked.


#9

the response I got when I put in a support ticket (my server is crunch) was that the issue they mentioned on the support site the other day about the Nailer server is still going on, and it’s causing abnormally high I/O traffic on all the servers it connects to, which slows them to a crawl. The tech said they’d be installing some new hardware wednesday night to fix it.


#10

Peeking at other stupid users’ processes that is.