On Friday 05 November 2010 06:52:31 ltsp-discuss-requ...@lists.sourceforge.net
wrote:
> > Hi everyone
> >
> > I'm in the process to update the server of one of my customers. It is a
> > small non-profit company.
> >
> > Their current server is an old Intel 2,4GHz P4/Celeron (one core) with
> > 512 Mb of Ram and 3 IDE hard disks in Raid 6. It is setup with LTSP 5.1
> > on Debian Lenny and a KDE 3.5.10 desktop.
> > There are 6 users on this server (it can grow to 10 users for short
> > periodes), using mainly kmail, konqueror, flash and OpenOffice.org.
> > Occasionally, they need to burn some CD and to play DVDs. They use the
> > current server to do that.
> > Thin clients are Artec ThinCan DBE62 (Atom CPU, 256 Mb Ram) and old PII
> > computers.
> >
[snip]
> A SSD for the OS is nice but don't skimp here. Either get a nice one
> or don't bother. A NAS for the home dis would be fine but don't force
> squid to cache over the network, it will be a pretty big bottleneck.
> Just buy a dedicated disk for squid, with 1 TB drives for less then
> $100 it's hard to go wrong and you don't really care if it dies. Just
> drop another one in and wait for squid to re-populate. In fact the
> quid guys recommend NOT using a RAIDed disk.
>
> > - where should I put the virtualized Windows? SSD or NAS?
> > - same question with the /tmp partitions.
>
> If I were you I'd put the windows virtual hard drive and on the SSD.
> /tmp should live there for sure. It gets hit pretty hard when you use
> LTSP.
Unrelated to ltsp, I found and used these for ssd options:
------- ssd changes
#!/bin/bash
# must be run with sudo
# setup changes for SSD
# /usr/lib/pm-utils/power.d/powersave-policy-dirty-writeback
# modify /etc/sysctl.conf to change kernel dirty page writeback time
test=`grep '\# I added to reduce disk activity' /etc/sysctl.conf`
if [ "$test" = "" ]; then
if [ ! -e /etc/sysctl.conf-orig ]; then
cp -a /etc/sysctl.conf sysctl.conf-orig
fi
echo >> /etc/sysctl.conf
echo '# I added to reduce disk activity to 240 seconds' >>
/etc/sysctl.conf
echo 'vm.dirty_ratio = 40' >> /etc/sysctl.conf
echo 'vm.dirty_background_ratio = 1' >> /etc/sysctl.conf
echo 'vm.dirty_writeback_centisecs = 24000' >> /etc/sysctl.conf
fi
# Note: if you DON'T have a laptop, uncomment the line below
# to prevent changes possibly being undone
chmod -x /usr/lib/pm-utils/power.d/laptop-mode
# edit fstab to mount ramdrive to /tmp
test=`grep '^tmpfs.*\/tmp.*tmpfs' /etc/fstab`
if [ "$test" = "" ]; then
if [ ! -e /etc/fstab-orig ]; then
cp -a /etc/fstab /etc/fstab-orig
fi
echo >> /etc/fstab
echo '# I added to reduce disk activity' >> /etc/fstab
echo 'tmpfs /tmp tmpfs defaults,noatime,size=1000M,mode=1777 0 0' >>
/etc/fstab
echo '# The following lines are disabled because of the bug in
mountall'
>> /etc/fstab
echo '# These may be mounted in /etc/rc.local instead' >> /etc/fstab
echo '# See
https://bugs.launchpad.net/ubuntu/+source/mountall/+bug/431040' >> /etc/fstab
echo '#tmpfs /var/log tmpfs defaults,noatime,size=200M,mode=0755 0 0'
>>
/etc/fstab
echo '#tmpfs /var/tmp tmpfs defaults,noatime,size=500M,mode=1777 0 0'
>>
/etc/fstab
echo '#tmpfs /var/spool/postfix/public tmpfs
defaults,noatime,size=50K,mode=1777 0 0' >> /etc/fstab
fi
# Replace rc.local for SSD changes
if [ ! -e /etc/rc.local-orig ]; then
cp -a /etc/rc.local /etc/rc.local-orig
fi
cat << EOF > /etc/rc.local
#!/bin/bash
# above must be #!/bin/bash not default /bin/sh -e
#
# rc.local
#
# This script is executed at the end of each multiuser runlevel.
# Make sure that the script will "exit 0" on success or any other
# value on error.
#
# In order to enable or disable this script just change the execution
# bits.
#
# By default this script does nothing.
# mount ramdrives to /var/tmp and /var/spool/postfix/public
for dir in "/var/tmp" "/var/spool/postfix/public" ; do
test=`mount | grep " on $dir "`
if [ "$test" = "" ]; then
sz="50K"
md="1777"
case "$dir" in
/var/tmp )
sz="500M"
md="1777"
;;
/var/spool/postfix/public )
sz="50K"
md="1777"
;;
esac
mount -t tmpfs -o size=$sz,noatime,mode=$md tmpfs $dir
fi
done
# Set sda to use deadline scheduler
# Note: if these changes don't seem to take, try pasting them into a root
console
# (or with sudo) manually once then reboot
echo deadline > /sys/block/sda/queue/scheduler
echo 1 > /sys/block/sda/queue/iosched/fifo_batch
exit 0
EOF
chmod u+x /etc/rc.local
# If you run postfix, change postfix startup script so that rc.local runs first
# Note: This may need to be repeated if postfix is upgraded
test=`grep 'bash /etc/rc.local' /etc/init.d/postfix`
#if [ "$test" = "" ]; then
# sed -i -n '
# 1h
# 1!H
# $ {
# g
# s/case "$1" in\x0A start)\x0A/case "$1" in\x0A
start)\x0Abash
\/etc\/rc.local\x0A/
# p
# }
# ' /etc/init.d/postfix;
#fi
exit
----- ssd test
#!/bin/bash
# Test SSD settings
echo
'###########################################################################'
test=`cat /proc/mounts | grep "^/dev/sda1.*commit="`
if [ "$test" = "" ]; then
echo "BAD: $test"
fi
test=`cat /proc/mounts | grep "^/dev/sda2.*commit="`
if [ "$test" = "" ]; then
echo "BAD: $test"
fi
test=`cat /proc/sys/vm/dirty_ratio`
if [ "$test" != "40" ]; then
echo "BAD: /proc/sys/vm/dirty_ratio $test"
fi
test=`cat /proc/sys/vm/dirty_background_ratio`
if [ "$test" != "1" ]; then
echo "BAD: /proc/sys/vm/dirty_background_ratio $test"
fi
test=`cat /proc/sys/vm/dirty_writeback_centisecs`
if [ "$test" != "24000" ]; then
echo "BAD: /proc/sys/vm/dirty_writeback_centisecs $test"
fi
test=`cat /sys/block/sda/queue/scheduler | grep "[deadline]"`
if [ "$test" = "" ]; then
echo "BAD: /sys/block/sda/queue/scheduler $test"
fi
test=`cat /sys/block/sda/queue/iosched/fifo_batch`
if [ "$test" != "1" ]; then
echo "BAD: /sys/block/sda/queue/iosched/fifo_batch $test"
fi
test=`mount | grep "tmpfs on /var/tmp "`
if [ "$test" = "" ]; then
echo "BAD: tmpfs on /var/tmp"
fi
test=`mount | grep "tmpfs on /tmp "`
if [ "$test" = "" ]; then
echo "BAD: tmpfs on /tmp"
fi
exit
James
------------------------------------------------------------------------------
The Next 800 Companies to Lead America's Growth: New Video Whitepaper
David G. Thomson, author of the best-selling book "Blueprint to a
Billion" shares his insights and actions to help propel your
business during the next growth cycle. Listen Now!
http://p.sf.net/sfu/SAP-dev2dev
_____________________________________________________________________
Ltsp-discuss mailing list. To un-subscribe, or change prefs, goto:
https://lists.sourceforge.net/lists/listinfo/ltsp-discuss
For additional LTSP help, try #ltsp channel on irc.freenode.net
