Virtual SME Server
SME Server as a virtualized Guest server
Considerations
- Storage: Local, NAS, iSCSI, LVM, Raid
- Network: LAN/WAN, VLAN, VPN, Bandwith,
- Out of band access (VNC, SPICE)
'Hardware' configuration of a Virtual SME Sever
CPU
- Host CPU or emulate
- Sockets and cores
Memory options
- To balloon or to not to balloon
Disk options
- Virtio driver or legacy driver?
- Disk types pros and cons
- Disk I/O options
Network options
- Virtio driver or legacy driver?
- Bridge, NAT or Route?
- Bandwidth options
Installation options of a Virtual SME Server
Kernel options
- LVM (option nolvm)
- Raid (option raid=none)
SME Server configuration settings
NTPD
Timing related options are important within Virtual Guests and to the amount of 'pressure' it puts on the host and level/increasing CPU usage of the host and guest. By default SME Server uses the NTP deamon for 'timing' related matters, but by default is focussed on the above mentioned 1000HZ, hence the kernel option 'divider=10', thus reducing the timing cycles/context switching requests on the host. See the above VMWare document mentioned (Way at the bottom).
On a Virtual SME guest server the ntpd SupportLargeDrift DB variable can be enabled as follows:
config setprop ntpd SupportLargeDrift enabled expand-template /etc/ntp.conf
and
service ntpd restart
to activate the new configuration. This will adjust /etc/ntp.conf to 'better' settings for a virtual guest. By setting the above value to 'disabled' and expand the template, the NTP service and configuration will revert back to SME Server defaults.
Clock/frequency
As per suggestions in this article from VmWare and this one from Oracle on virtual Linux Guests, adjusting the guest Frequency will improve the guests speed.
You can check if your guest server can benefit from these boot options:
[root@sme8 ~]# grep CONFIG_HZ /boot/config-`uname -r` # CONFIG_HZ_100 is not set # CONFIG_HZ_250 is not set CONFIG_HZ_1000=y CONFIG_HZ=1000
If you see the above result, these boot options are useful.
For example, if your kernel boot line is:
kernel /vmlinuz-2.6.18-348.6.1.el5 ro root=/dev/main/root
change it to:
kernel /vmlinuz-2.6.18-348.6.1.el5 ro root=/dev/main/root divider=10 clocksource=acpi_pm
Tools/utilities
iotop
top like utility to monitor the guest I/O (performance)
yum install iotop
cpu info
Detailed info on the guest cpu(s)
cat /proc/cpuinfo
Proxmox
- Interesting article on installing SME Server as guest on Proxmox
SME v8 on Proxmox at Online.net
This is based on work by Daniel here: https://wikit.firewall-services.com/doku.php/tuto/virtualisation/netwok_conf_sur_dedibox
Online.net configuration
Some assumptions:
- Your main IP address
- IP address 62.20.20.250
- Netmask 255.255.255.0
- Gateway 62.20.20.1
- Your secondary IP address
- 200.30.30.1
- 10:10:00:00:20:20 - you DID set the MAC didn't you ?
Proxmox configuration
We will use the vmbr0 that is set up, but we need to create a dummy one so we can put SME in server/gateway mode and use Daniels scripts:
- Create a new network interface and call it vmbr1
- Do NOT set IP address/subnet/bridge ports etc.
- Create your KVM for SME. Make sure that you give it two network adaptors - one is vmbr0 and one is vmbr1
- vmbr0 should have the following settings :
- Model: virtio
- MAC: 10:10:00:00:20:20 (as per your mac that you set)
- After you have created the KVM add a second adaptor:
- vmbr1 should have the following settings:
- Model: virtio
- MAC: should be the MAC of your main IP address
It doesn't matter which one is 0 or 1 as long as the settings are correct.
SME Configuration
Now do your standard SME installation.
<here would be a lot easier if we could force it to have a local IP first so you could ssh from Proxmox to SME and you could then use scp to copy files or a terminal to copy and paste>
When you go through the setup, make sure you pick the DUMMY adaptor for the local interface, and the real adaptor for the external interface. Make sure you know your MAC addresses so you know which one is which.
you can use your second IP address for the external interface, but you will have to set a subnet mask and gateway IP that it likes e.g.:
- IP : 200.30.30.1
- Subnet : 255.255.255.0
- Gateway : 200.30.30.1
One you have rebooted your network will not be working correctly so we need to follow Daniels guide to fix it :
db configuration set ExternalIP 200.30.30.1 db configuration set ExternalNetmask 255.255.255.255 db configuration set GatewayIP 62.20.20.1 /etc/e-smith/events/actions/initialize-default-databases
mkdir -p /etc/e-smith/templates-custom/etc/sysconfig/network/
echo '# GATEWAYDEV disabled for dedibox network' > /etc/e-smith/templates-custom/etc/sysconfig/network/40GATEWAYDEV
echo '# GATEWAY disabled for dedibox network' > /etc/e-smith/templates-custom/etc/sysconfig/network/50GATEWAY
expand-template /etc/sysconfig/network
mkdir -p /etc/e-smith/templates-custom/etc/sysconfig/network-scripts/route-ethX
cat <<'EOF' > /etc/e-smith/templates-custom/etc/sysconfig/network-scripts/route-ethX/20gateway
{
die "Need to pass THIS_DEVICE in MORE_DATA\n" unless (defined $THIS_DEVICE);
my $device = $ExternalInterface{Name};
unless ($device)
{
warn("Can't determine device name for external network");
return "# template expansion error - Can't determine device name for external network";
}
return "# Gateway only applies on external interface"
unless ($ExternalInterface{Name} eq $THIS_DEVICE);
my $gw = $ExternalInterface{'Gateway'};
$OUT .= "$gw dev $THIS_DEVICE\n";
$OUT .= "default via $gw dev $THIS_DEVICE";
}
EOF
If you have got it right then you should be able to run the following without error :
expand-template /etc/sysconfig/network-scripts/ifcfg-$(db configuration getprop ExternalInterface Name) expand-template /etc/sysconfig/network-scripts/route-$(db configuration getprop ExternalInterface Name)
Check all the network settings look correct and then:
signal-event post-upgrade;signal-event reboot
You should now be able to connect to your Proxmox box on your primary IP address, and your SME server on your added IP.
Your network setting should look similar to this:
EthernetDriver1=virtio_net EthernetDriver2=virtio_net ExternalDHCP=off ExternalIP=200.30.30.1 ExternalInterface=interface Broadcast=(whatever) Configuration=static Driver=virtio_net Gateway=62.20.20.1 #gateway for your main IP address HWAddress=10:10:00:00:20:20 #MAC that you set for 2nd IP address IPAddress=200.30.30.1 #your 2nd IP address Name=eth1 Netmask=255.255.255.255 Network=200.30.30.1 #your 2nd IP address ExternalNetmask=255.255.255.255 GatewayIP=62.20.20.1 #gateway for your main IP address
Following your 'local network' and is the range when you VPN :
InternalInterface=interface Broadcast=192.168.98.255 Configuration=static Driver=virtio_net HWAddress=fe:aa:16:19:e8:bf #MAC for 'virtual' adaptor IPAddress=192.168.98.1 NICBondingOptions=miimon=200 mode=active-backup Name=eth0 Netmask=255.255.255.0 Network=192.168.98.0 LocalIP=192.168.98.1 LocalNetmask=255.255.255.0
Xen
QEMU/KVM
Virtualbox
High I/O causing filesystem corruption
It seems during times of high I/O, like backups and etc, the guest filesystem resets to read-only. You can have the following errors in log
https://www.virtualbox.org/ticket/10031
Apr 30 13:37:29 sme9 kernel: ata2.00: exception Emask 0x0 SAct 0x7fffffff SErr 0x0 action 0x6 frozen Apr 30 13:37:29 sme9 kernel: ata2.00: failed command: WRITE FPDMA QUEUED Apr 30 13:37:29 sme9 kernel: ata2.00: cmd 61/08:00:08:7d:77/00:00:03:00:00/40 tag 0 ncq 4096 out Apr 30 13:37:29 sme9 kernel: res 40/00:01:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
the workaround is to set virtualbox to use the host I/O disk cache : see settings/storage/controller and activate the option
SME Server as a Host server
Phpvirtualbox
- Phpvirtualbox SME Contrib