For the techies that like dropbox a simple way also to include your ubuntu karmic server.

Login with SSH to your server.

Put these rules at your server in the /etc/apt/sources.list

deb http://linux.dropbox.com/ubuntu karmic main
deb-src http://linux.dropbox.com/ubuntu karmic main

On your server do an aptitude update && aptitude install nautilus-dropbox

Then answer Yes to all… and it will install.

Logout from your terminal and you must be from a linux / x-based workstation to complete the install.

Login back with the following command:

ssh -XY username@yourserver.com

Now you have logged in with a terminal X-Session. Now run the following command:

dropbox start -i –sync

Your dropbox window will appear on the screen and just follow installation as normal. It may differ from the real world experience, but just continue.

If the install is ready close all windows, yes also your X-Term towards the server.

And yet again login normally with ssh yourname@yourserver, and run the following command:

dropbox start &

Now you have a dropbox folder that works on your server 24/7.

Instead of running dropbox with & you can use the solution found in this article: http://kevin.vanzonneveld.net/

Ok now some solaris handy’s that I need to remember while I am testing my home server setup:

pkg install SUNWgcc ss-dev

That will install GCC and development tools for my machine. So I could install smartmon tools (get smartmon, ./configure; make; make install works!)

# cat /etc/power.conf

device-dependency-property removable-media /dev/fb
autopm enable
autoS3 default
S3-support enable
autoshutdown 30 9:00 9:00 noshutdown
cpu-threshold 10s
cpupm enable poll-mode
device-thresholds /pci@0,0/pci1043,82d4@1f,2/disk@1,0 1200s
device-thresholds /pci@0,0/pci1043,82d4@1f,2/disk@2,0 1200s
device-thresholds /pci@0,0/pci1043,82d4@1f,2/disk@3,0 1200s
device-thresholds /pci@0,0/pci1043,82d4@1f,2/disk@4,0 1200s

That will scale my CPU to a lower profile when he has nothing to do, and also enables my four harddrives to spin down when they are idle.

At this moment the current configuration looks like this:

Intel E7500 Core 2 Duo processor
Asus PQ5 premium
4 * 1 TB 7200RPM samsung F1 drives
Intel bootup SSD 40 gigabytes X-25-V
Intel cache SSD 32 gigabytes X-25-E
6 gigabyte DDR-2 memory.
1 LSI PCI-E x8 2 port SAS controller.

In backorder to have a dedicated zil is a gigabyte I-RAM! Have to wait for this, until now it’s on a 4 gigabyte partition on the Intel X-25-E SSD.

I had the two seagate’s but two disks did not perform the way I would like, and also the disks where 5900rpm and the 4 new disks are 7200rpm. The 80 gigabyte Intel X-25-M I also removed as I have many writes also on my platform and I could see that with every write the L2ARC also writes some stuff to the Intel X-25-M, and that’s why I am shipping in a gigabyte I-RAM, and replace the Intel X-25-M with the Intel X-25-E, and use the I-RAM for the ZIL.

Also the slow CPU was removed due to the CPU load by deduplication and compression. With the faster CPU and the disk changes (and even without the I-RAM) I now get about 280mb/s writes with a 16 gigabyte file created with “dd”, and around 640mb/s of reads when doing it the other way around. With the slow cpu and the other harddrives I was getting around 80mb/s writes and 110mb/s of reads. So a big performance increase. When the I-RAM is in the X-25-E will be relieved of the ZIL log tasks so that would increase NFS/I-SCSI performance a bit when several stations are reading and writing at the same time.

At this moment in time I am running Opensolaris version 133 from http://genunix.org one problem though is that the second internal sata controller (sil based / marvell based) does not work yet. But the 6 ports from the ICHR-10 chipset works great and also the extra lsi controller works perfectly (that will be used for the l2arc device and the zil device). I will update to 134 any time soon.

Also some cool stuff about drive temp and other zfs know abouts you find at http://breden.org.uk/2008/03/02/a-home-fileserver-using-zfs/

Ok I am writing this article also for me as a reminder of what commands to use and what are the best options. At this point in time I have the plan to build a home NAS system with ISCSI, NFS, CIFS (samba) functionallity and completely stop the usage of harddrives in my local workstation network that is now 2 linux desktop machines and 2 windows XP machines. Also I want a good and reliable storage device that protects the data on it.

So in a few simple words I want a nas that can automate the process for each workstation.

For a first remark to boot ISCSI devices from the network without buying expensive network cards you can check out http://www.etherboot.org this site offers the most functionallity to boot from your network with every kind of operating system.

What I already worked out is that I need a DHCP server that can boot images for PXE, all my pc’s at home support gigabit ethernet PXE, so I can easily feed the workstation the correct gPXE rom and that will boot ISCSI targets from the mainserver.

The mainserver I have ordered is a simple Intel core 2 Duo E2140 cpu that supports 64-bit extensions, this cpu will be run on an Asus Pq5-premium motherboard with 10 sata-II and 4 gigabit ethernet ports. The harddrives are 2 seagate baracuda’s 5900rpm 1.5TB devices that will in mirror. The cache (l2arc) and the log (zil) device will be an Intel X-25-M 80 gigabyte and a Intel X-25-E 32 gigabyte. Equiped with 6 gigabytes of ram and installed with the latest build (version 133) from http://www.genunix.org it should be a cool and fast device with an eye towards the future if I need more disk space or a faster cpu (quad core e.g.) or more memory, or even more bandwidth on my local network.

Ok before planning this I had to figure out what Opensolaris is, and what I have been reading about ZFS was true. So this weekend I pulled out a small dell optiplex with a simple p4 2.4ghz processor and 1,2GB of ram and a small IDE 40 gigabyte disk. And installed opensolaris on it the latest build ofcourse because I wanted a cool thing that people called deduplication, just like our netapp in the colo does and snapshots are not bad either to have.

Here are some commands i’ve used, as always I won’t give a lot of text, you just have to figure that out it are rough guidelines.

- format (leave with control-c format as root will give you all information you need in what kind of disks you have.)

- zpool create tank mirror c4d0p1 c4d1p1 (where c4d0p1 and the other are your hd’s, this creates a mirror from two of the devices)

- zpool add tank log c4d2p1 (this will be the 32gb ssd as the ZIL / LOG device this will give a write performance boost as stated in other documentation.)

- zpool add tank cache c4d3p1 (this will be the 80 gigabyte CACHE / L2ARC device that will give read performance.)

So now we have created a /tank pool that has maximum write and read performance. Don’t ask me how much performance it will give since I don’t have the hardware yet and can’t test much about it, but if you google on L2arc and Zil with ZFS and Opensolaris you can find a lot of documents giving performance figures and they are outstanding!

For the people that need to ask, I have tested this on my small pc on the same drive in different partitions and that really sucks, but I have to wait until the good stuff is delivered.

Now let’s fire up the compression and the deduplication.

- zfs set compression=on tank
- zfs set dedup=on tank

For a little example I have put my /home/dennis directory from my local linux pc on that newly created drive. My directory is about 9.6 gigabytes and houses are sort of stuff, and the compression brought it down to about 7.4 gigabytes and the deduplication (rules!) brought it down to 6.2 gigabytes! So that will save a lot of space considering the stuff I will put on it.

So since I am new to Opensolaris and it’s command structure I figured out how to get a NFS share that I need to share documents between the two linux hosts and the windows hosts

- zfs create tank/share
- zfs sharenfs=on tank/share

Allow only one host:
- zfs sharenfs=rw=@192.168.1.119,root=@192.168.1.119 tank/share

Allow all the hosts on your subnet:
- zfs sharenfs=rw=@192.168.1.0/24,root=@192.168.1.0/24 tank/share

Ofcourse windows hosts don’t like NFS so we need to use CIFS (samba) to get them to our share.

- pkg install SUNWsmbs
- pkg install SUNWsmbskr
- svcadm enable -r smb/server
- zfs sharesmb=on tank/share
- zfs sharesmb=rw=@192.168.1.0/24 tank/share
- vi /etc/pam.conf and add the following line:
other password required pam_smb_passwd.so.1 nowarn
- create a local user that you will use to access the windows share

Now you can access with that user the share from your windows computer.

If you want to call your share something like COOLSHARE only because it is possible:

- zfs set sharesmb=name=COOLSHARE tank/share

And if you want to join a workgroup with your solaris server then just type:

- smbadm join -w YourWorkGroup

Since I wanted to etherboot all workstations with ISCSI i needed also to support ISCSI, a good thing is that all my computers are already installed so I will “dd” the complete drive towards a image and set it up in an ISCSI share, but for now just let’s enable it on the opensolaris server.

- pkg install SUNWiscsitgt
- svcadm enable iscsitgt
- zfs create -V 5G tank/iscsi
- zfs set shareiscsi=on tank/iscsi
- iscsitadm create target -b /dev/zvol/rdsk/tank/iscsi iscsi
- iscsitadm list target -v

How cool and simple is that? To show that it works I had to install a package on my ubuntu workstation:

- aptitude install open-iscsi
- iscsiadm -m discovery -t sendtargets -p 192.168.1.136 (.136 is
the opensolaris server and returns the following:
192.168.1.136:3260,1 iqn.1986-03.com.sun:02:40cd2097-db92-ce54-8864-84e64ff4a5a4

And maybe some errors with IPV6, but you can ignore those.
)

- iscsiadm -m node –targetname iqn.1986-03.com.sun:02:40cd2097-db92-ce54-8864-84e64ff4a5a4 -p 192.168.1.136:3260 –login

And ….

[49773.143342] scsi5 : iSCSI Initiator over TCP/IP
[49773.440133] scsi 5:0:0:0: Direct-Access SUN SOLARIS 1 PQ: 0 ANSI: 5
[49773.440375] sd 5:0:0:0: Attached scsi generic sg2 type 0
[49773.441179] sd 5:0:0:0: [sdb] 10485760 512-byte logical blocks: (5.36 GB/5.00 GiB)
[49773.447472] sd 5:0:0:0: [sdb] Write Protect is off
[49773.447482] sd 5:0:0:0: [sdb] Mode Sense: 67 00 00 08
[49773.449828] sd 5:0:0:0: [sdb] Write cache: disabled, read cache:
enabled, doesn't support DPO or FUA
[49773.452632] sdb: unknown partition table
[49773.461441] sd 5:0:0:0: [sdb] Attached SCSI disk


So that’s all for now, what I have to do now is wait for my hardware to arrive, install the machine and the software, and make some tests with booting over ISCSI and sharing all the stuff between the workstation all from opensolaris.

What I have been reading about is that opensolaris uses a different DHCP server then the DHCP3D server from linux, so maybe I will install virtualbox on the home server and setup a ubuntu server to take care of the PXE booting/addressing stuff. I will keep you notified.

Owh I was looking how many memory there was in use on the pc, you can find it with the commando top, or even cooler try this:

- echo “::memstat” | mdb -k

Hi again,

Today I am writing again on my blog, and finally some tech stuff is being added, just posted the wireless stuff and how to add extra ssid’s and put them in seperate vlan’s… Now we want the sales and tech department also being deported in different vlans if they login with PPTP from there home. So I took the cisco 870 and configured the following yet again I am not going to explain every detail, the config speaks for itself.


no service pad
service timestamps debug datetime msec localtime show-timezone
service timestamps log datetime msec localtime show-timezone
service password-encryption
service internal
service sequence-numbers
!
hostname ciscovpnbox
!
boot-start-marker
boot-end-marker
!
logging userinfo
logging buffered 51200 debugging
enable secret 5 blabla
enable password 7 blabla
!
aaa new-model
!
!
aaa authentication ppp default local
aaa authorization network default local
!
aaa attribute list sales
attribute type addr-pool "sales"
!
aaa attribute list tech
attribute type addr-pool "tech"
!
aaa session-id common
!
resource policy
!
ip subnet-zero
ip host-routing
no ip gratuitous-arps
ip cef
!
!
ip dhcp smart-relay
ip dhcp bootp ignore
!
!
no ip bootp server
ip domain name ponne.nu
ip multicast-routing
ip ssh source-interface FastEthernet4
ip ssh version 2
ip address-pool local
vpdn enable
!
vpdn-group 1
! Default PPTP VPDN group
accept-dialin
protocol pptp
virtual-template 1
!
!
!
!
username techie privilege 15 password 7 bloediebloe
username techie aaa attribute list tech
username salesdude privilege 15 password 7 bladieblu
username salesdude aaa attribute list sales
!
!
!
!
!
!
interface FastEthernet0
switchport trunk allowed vlan 1,2,10,11,1002-1005
switchport mode trunk
spanning-tree portfast
hold-queue 100 out
!
interface FastEthernet1
!
interface FastEthernet2
!
interface FastEthernet3
!
interface FastEthernet4
ip address "externalipaddress"
duplex auto
speed auto
!
interface Virtual-Template1
ip unnumbered Vlan10
ip mroute-cache
no keepalive
ppp encrypt mppe auto passive
ppp authentication ms-chap ms-chap-v2
!
interface Virtual-Dot11Radio1
no ip address
!
interface Virtual-Dot11Radio2
no ip address
!
interface Vlan1
no ip address
!
interface Vlan10
ip address 192.168.0.200 255.255.255.0
!
interface Vlan11
ip address 192.168.1.200 255.255.255.0
!
ip local pool tech 192.168.0.201 192.168.0.211
ip local pool sales 192.168.1.201 192.168.1.211
ip classless
ip route 0.0.0.0 0.0.0.0 "ipaddress"
!
!
no ip http server
no ip http secure-server
!
logging trap debugging
logging source-interface FastEthernet4
no cdp run
!
!
control-plane
!
!
line con 0
no modem enable
line aux 0
line vty 0 4
password 7 boedoebloek
transport input ssh
!
scheduler max-task-time 5000
scheduler allocate 4000 1000
scheduler interval 500
end


With this you can have those two users login separately from each other in there own vlan how cool is that with a simple VPN box!