I have a problem running a printserver on top of VMware ESX.
The machine is running with a very high cpu-load.
First I want to describe the installation:
Hostname: cobu0035
OS: SLES 10 SP1
Kernel: Linux version 2.6.16.54-0.2.10-smp
VM-Tools: Version 3.5.0, build-123630
CPU's: Log: HOSTINFO: Seeing Intel CPU, numCoresPerCPU 4 numThreadsPerCore 1.
Log: HOSTINFO: numPhysCPUs is 0, bumping to 1.
Log: HOSTINFO: This machine has 1 physical CPUS, 2 total cores, and 2 logical CPUs.
bootparameter:
root=/dev/system/lvroot resume=/dev/system/lvswap splash=silent showopts clock=notsc
Application:
The machine is running a cups Printserver Version: cups-1.1.23-40.46
And a Samba Fileserver as frontend to host the printerdrivers.
The server is running very well. The machines is hosting 220
printers with approx. 5000-6000 printjobs per day.
Problem:
The machines seems to be very slow, when somebody try to connect
to the samba-shares.
To track down this issue I have done the following steps:
When I do a uptime I got the following output:
cobu0035:/usr/lib/vmware-tools/bin32 # uptime
11:39am up 4 days 18:35, 2 users, load average: 1.75, 1.97, 1.88
The load average seems to be very high!
When I do a top I got this output:
cobu0035:/usr/lib/vmware-tools/bin32 # top
top - 11:41:13 up 4 days, 18:36, 2 users, load average: 1.57, 1.83, 1.83
Tasks: 337 total, 5 running, 332 sleeping, 0 stopped, 0 zombie
Cpu0 : 19.4%us, 23.7%sy, 0.0%ni, 55.6%id, 0.3%wa, 0.0%hi, 1.0%si, 0.0%st
Cpu1 : 41.3%us, 20.1%sy, 0.0%ni, 36.3%id, 0.0%wa, 1.0%hi, 1.3%si, 0.0%st
Mem: 2075592k total, 1563360k used, 512232k free, 95196k buffers
Swap: 1048568k total, 108k used, 1048460k free, 192624k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
19137 root 18 0 99108 92m 1568 R 33 4.5 305:04.11 cupsd
9431 BROSE+lo 15 0 44808 9160 4452 S 26 0.4 0:05.16 smbd
10759 BROSE+zi 15 0 49972 10m 6420 S 20 0.5 0:01.59 smbd
3140 root 15 0 55344 10m 8584 S 6 0.5 128:09.84 winbindd
When we look into the two lines with Cpu0 and Cpu1 there is a very high
systmload. This means the application cupsd does a lot of systemcalls.
I followed the process via strace and got a
output like this:
...
time(NULL) = 1226659685
time(NULL) = 1226659685
time(NULL) = 1226659685
select(1024, [0 2 3 5 7 8 9 10 11 12 13 14 15 16 17 18 19 20 22 24 25 26 27
28 29 30 31 32 33 34 35 36 38 39 42 43 44 45 46 47 48 49 50 51 52 53 54 55
56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 78 79 80 81
82 83 84 85 86 87 88 89 90 91 92 95], 22 24 26 28 35 38, NULL, {1, 0}) = 6
(out 22 24 26 28 35 38, left {1, 0})
time(NULL) = 1226659685
time(NULL) = 1226659685
time(NULL) = 1226659685
time(NULL) = 1226659685
time(NULL) = 1226659685
time(NULL) = 1226659685
time(NULL) = 1226659685
time(NULL) = 1226659685
time(NULL) = 1226659685
time(NULL) = 1226659685
....
this is repeating and repeating.
It looks very boring, but this means the software is always asking about the
time!
Do I have any chances to speedup this calles?
I know it is wellknow that VMware often has problem with timesync, so we added the clock=notsc but no change.
Any Ideas?
If you found this or any other answer useful please consider the use of the Helpful or correct buttons to award points
