> Subject: Re: [K12OSN] LTSP CentOS5 slow
> From: brcisna eazylivin net
> To: k12osn redhat com
> Date: Fri, 28 Nov 2008 08:23:00 -0600
> Hi Jones,
> Please supply some specs of your server and how many TC's you are
> connecting to it.Is this a default install of K12LTSP5, or CentOS5 with
> ltsp on top of it? Have you updated the server since a default install?
> The FF update shouldn't have hindered the performance any just,, FYI.
> A few things to try.
> 1. run top from a terminal while the server seems " slow ", and see if
> your memory is all used up,or if cpu is 100%.
> 2. You could download jperf,and install on your K12LTSP server as well
> as a (linux)laptop you have available. Plug your laptop into the switch
> that the TC's are plugged into. Run jpef in server mode on the server &
> run jperf on your laptop in client mode. You will see what bandwidth is
> available along with jitter and all that sorta geeky stuff.This piece
> has really helped me in finding were bandwidth is being used realtime.
> 3. Is the switch the TC's are plugged into a 'plug and play' switch? In
> other words a non-managed switched?
> 4. That is odd Firefox is throwing that error you posted. Did you
> install FF via yum? Is this error happening every time FF is launched,
> or hit and miss?
> You can download the jperf rpm at the URL below. You will need to
> install iperf, first, which is at the url as well.
> Let us know your findings.
> Take Care,
> Barry Cisna
> K12OSN mailing list
> K12OSN redhat com
> For more info see <http://www.k12os.org>
I wasn't able to find the loopback and I'm curious if the switch has green lights because of the possible Wake-On-LAN abilities of the NIC cards. The NIC lights are green when the computer is plugged in but turned off.
Currently all 30 TC's are off and its just 1 TC and 1 server.
To answer your previous questions:
There are max 31 thin clients.
2 15K rpm scsi drives in software raid 1
6GB ram (serving max 30 students at a time)
1. TOP output:
top - 16:40:27 up 4 days, 2:10, 7 users, load average: 0.16, 0.24, 0.20
Tasks: 211 total, 1 running, 208 sleeping, 0 stopped, 2 zombie
Cpu(s): 2.9%us, 0.2%sy, 0.0%ni, 96.6%id, 0.2%wa, 0.0%hi, 0.1%si, 0.0%st
Mem: 3631812k total, 2337164k used, 1294648k free, 357848k buffers
Swap: 1036088k total, 160k used, 1035928k free, 1278744k cached
[I was tired earlier and Swapped Mem and Swap] Memory looks like its being used up. This is with just 1 TC running.
I have both window and Linux in my lab, after reading I noticed that I could install iperf.exe on a Windows client and I won't need a laptop. I had key issues installing iperf on Linux, but it says it installed it. I installed jperf no problem, but when I tried to run it I got this:
Error occurred during initialization of VM
Could not reserve enough space for object heap
Could not create the Java virtual machine.
3.I believe that the switch is non-managed. We just connected the cables in and away we went.
4. Yes, I installed Firefox with yum. I also updated a couple of times. I updated it again today. Firefox crashes every once in a while. It was crashing more regularly before the update.
It seems to be fine when I turn the switch off/on. If its just a couple of users then it seems to be running ok. Once a class gets in its slow again.
I'm stuck again.