Hetzner - DokuWiki

VMware ESXi/en
(Hardware)
(Updating vSphere 5.0 to 5.1)
(84 dazwischenliegende Versionen von 11 Benutzern werden nicht angezeigt)
Zeile 1: Zeile 1:
{{Languages|VMware ESXi}} <br>
+
{{Languages|VMware ESXi}}
= General information  =
+
  
== Physical Hard Disks  ==
+
= General information =
  
* VMware ESXi uses a filesystem called vmfs5 (formerly vmfs3), which uses GPT and supports disk sizes of up to 64 TiB (formerly 2 TiB).
+
== Hardware ==
* Software RAID (as used in our regular root servers) is not possible, only hardware RAID controllers can be used.
+
  
== Networking  ==
+
*VMware vSphere Hypervisor 5.x and 6.0 (ESXi) use a filesystem called vmfs5 (formerly vmfs3), which uses GPT and supports drive sizes of up to 64 TiB (formerly 2 TiB).
 +
*Software RAID is *not* supported by ESXi.
 +
*The free version of ESXi starting with 5.5 supports a maximum of 4 TB RAM (formerly 32 GB RAM).
  
*ESXi is an operating system designed for virtualization. Its networking features do not include NAT or routing.
+
== Network configuration ==
*Additional IP addresses can only be used in bridge mode, which requires additional MAC addresses to be assigned. To use a subnet one of the additional IP addresses has to be setup as a router VM.
+
  
== Installation  ==
+
*VMware vSphere Hypervisor is an "operating system" designed for pure virtualization and supports neither NAT nor routing. Therefore, only a true bridged setup can be used.
 +
*To use a subnet an additional single IP needs to be setup as a router VM.
  
*The installation and basic configuration of ESXi with the install disc takes about 20-30 minutes, even without any prior knowledge.([[#Installation_guide|install guide]]).
+
== Installation ==
  
= Hetzner specific information =
+
*The installation and configuration of ESXi takes about 20-30 minutes, even without any prior knowledge. Please check out the [[#Installation Guide|Installation Guide]] for further information on installing ESXi.
  
== Hardware  ==
+
== License ==
  
*The EX and EQ series are fully compatible with ESXi (installation of ESXi on the older DS series is '''not''' possible)
+
*After installation the system has a test license, which expires after 60 days. For the long-term usage of the OS a separate (free) license is required. This can be requested by registering on the VMware website:
*When ordering your EQ server, please state in the comments section of the order form that you want to install ESXi. Hetzner Support will then provide you with a CD and a remote console (LARA) ready to install ESXi.
+
**License for [https://my.vmware.com/web/vmware/evalcenter?p=free-esxi5 ESXi 5.x]
*When installing on a server with multiple identical disks, ESXi can potentially show the disks in a different order than the BIOS does. Should the screen show only a blinking cursor after the installation is done, you should try the other disk as the boot disk.
+
**License for [https://my.vmware.com/web/vmware/evalcenter?p=free-esxi6 ESXi 6.0]
 +
*The license can be entered via the vSphere Client, under the tab "Configuration", the section "Software" and then under the menu item "Licensed functions". After selecting this you can click on the "Edit" button to the right and then enter the license key. [[File:Lizenzzuweisung_mit_vSphereClient.jpg|600px|Entering license with vSphere Client]]
  
== Network Setup ==
+
= Hetzner specific information =
  
*In our data centers IP addresses are statically mapped to the MAC addresses of servers. Thus it is necessary to request a special MAC address for the additional IP address because only bridge mode is possible. These MAC addresses have to be configured for a given virtual server and its IP address. To get these MAC addresses assigned, log in to Robot, choose "Server" from the menu, select the desired server, and then click on the "IPs" tab. Here you can click on the icon next to the IP address to get a MAC address. The Flexi-Pack is a pre-requisite for this.
+
== Installed hardware ==
  
[[Datei:Esxi-mac-setzen.png]]
+
The Dell PowerEdge models DX150, DX151, DX290, DX291&nbsp; and DX141 are certified and compatible with all versions from 5.0 onwards. DX152 and DX292 are compatible from 6.5 onwards.
  
== Subnet ==
+
All other models are not certified by VMware, yet in most cases they can be run with VMware vSphere/ESXi.
  
*Before ordering an additional subnet, you need at least one additional IP for a Router VM since ESXI cannot route itself. When ordering a subnet include that you are using ESXi and ask to have the subnet routed to the additional IP adress. You do NOT get a separate MAC address for every IP address of the subnet.
+
=== Compatibility ===
  
The confirmation email of the subnets contains (for example) the following information:
+
(All information is subject to change)
  
Below you will find your additional IP addresses added to
+
{| class="wikitable" style="margin-left: 10px"
the server 192.168.13.156.<br>
+
|-
Please note that you can use the subnet only for this server.<br>
+
! Model
IP: 192.168.182.16 /28
+
! vSphere/ESXi Version
Mask: 255.255.255.240
+
|-
Broadcast: 192.168.182.31<br>
+
| AX50-SSD/AX60-SSD
Useable IP addresses:
+
| from 6.5a additional Intel NIC required
192.168.182.17 70 192.168.182.30
+
|-
 +
| AX41/AX51<br/>
 +
| from 6.7 Update 2<br/>(potentially an additional NIC required)
 +
|-
 +
| AX100
 +
| from 6.7
 +
|-
 +
| AX160
 +
| from 6.5 Update 2 / 6.7
 +
|-
 +
| PX92
 +
| from 6.5
 +
|-
 +
| PX91/PX121
 +
| from 5.5 Update 1
 +
|-
 +
| PX90/PX120
 +
| from 5.1
 +
|-
 +
| PX61
 +
| from 6.0 Update 2
 +
|-
 +
| PX62<br/>
 +
| from 6.5 Update 2 / 6.7 Update 1
 +
|-
 +
| PX60/PX70
 +
| from 5.5 Update 1
 +
|-
 +
| EX61<br/>
 +
| from 6.5
 +
|-
 +
| EX52/EX62<br/>
 +
| from 6.5 Update 2 / 6.7 Update 1
 +
|-
 +
| EX41/EX41S/EX51/EX42<br/>
 +
| from 5.5 Update 3 / 6.0 Update 2<br/>(potentially an additional NIC required)
 +
|-
 +
| EX40/EX60
 +
| 5.0 - 5.1 Update 2, since 5.5 additional NIC required
 +
|-
 +
| EX6/EX6S/EX8/EX8S
 +
| from 5.0
 +
|-
 +
| EX4/EX4S/EX10
 +
| from 5.0 (with additional NIC)
 +
|-
 +
| EQ4/EQ6/EQ8/EQ9
 +
| 5.0 - 5.1 Update 2, since 5.5 additional NIC required
 +
|}
  
In order to use these IP addresses in virtual machines, a "router VM" with an additional NIC in the new subnet is necessary. The subnet itself requires a new vSwitch in ESXi to which all VMs in the subnet will be connected.
+
'''NOTES''':
  
 +
*The models mentioned above should work with the appropriate version. However, newer versions are not necessarily compatible. To make sure please gather your own information.
  
'''Note'''
+
*The Realtek driver present in ESXi 5.0 to 5.1 Update 2 was just a Tech Demo that has not been updated. Therefore, for smooth operation a server with an Intel NIC (PX60, PX91, DX151 etc.) is recommended.
  
According to http://www.vmware.com/support/vsphere4/doc/vsp_esxi41_vc41_rel_notes.html
+
*The installation of vSphere 5.5 or newer on the EX40, EX60 and some EX41/EX51 models requires a custom installation ISO that includes drivers for the Realtek NIC ([http://www.bussink.ch/?p=1228 http://www.bussink.ch/?p=1228]). Alternatively, an additional compatible NIC can be installed. Please find the cost of an additional compatible NIC here: [[Root Server Hardware/en#Miscellaneous|Root Server Hardware]]
poor TCP performance can occur in traffic-forwarding virtual machines with LRO enabled. Some Linux modules cannot handle LRO-generated packets. As a result, having LRO enabled on a VMXNET 2 or VMXNET 3 device in a traffic forwarding virtual machine running a Linux guest operating system can cause poor TCP performance. LRO is enabled by default on these devices.
+
  
Workaround: In traffic-forwarding virtual machines running Linux guests, set the module load time parameter for the VMXNET 2 or VMXNET 3 Linux driver to include disable_lro=1.
+
*Installing ESXi on the older DS or X servers is '''not''' possible.
  
=== Preparations in the vSphere-Client ===
+
*When installing on a server with multiple identical drives, ESXi can potentially show the drives in a different order than the BIOS does. Should the screen be blank with only a blinking cursor after the installation is done and you restart, try booting from the other drive.
  
#create vSwitch (in the example the name: 'subnet' is used) [[Datei:Esxi-vswitch1.png|vSwitch-Teil 1]]<br><br>[[Datei:Esxi-vswitch2.png|vSwitch-Teil 2]]<br><br>[[Datei:Esxi-vswitch3.png|vSwitch-Teil 3]]<br><br>[[Datei:Esxi-vswitch4.png|vSwitch-Teil 4]]<br><br>
+
== Network Configuration ==
#add a second NIC to the router VM. connected network: subnet (the previously created vSwitch)<br><br>[[Datei:Esxi-router-nic.png]]<br><br>
+
#NIC of a virtual machine in the subnet. connected network:subnet
+
  
The networking overview should show the following:[[Datei:Esxi-subnet.png]]
+
*For connectivity between multiple ESXi servers within the same subnet, host routes via gateway are necessary due to network security restrictions.
 +
 
 +
host A
 +
esxcfg-route -a <IP Host B> 255.255.255.255 <Gateway IP>
 +
 
 +
host B
 +
esxcfg-route -a <IP Host A> 255.255.255.255 <Gateway IP>
 +
 
 +
== Single IP addresses ==
 +
 
 +
*IP addresses are, by default, statically mapped to the MAC address of the host. It is possible however to get separate MAC addresses for the additional single IPs via Robot. These can then be configured for the virtual machines. To get these MAC addresses assigned, log in to Robot, choose "Server" from the menu on the left, select the desired server, and then click on the "IPs" tab. Here you can click on the icon next to the single IP address to get a MAC address.
 +
 
 +
[[File:Esxi-mac-setzen.png|600px|Esxi-mac-setzen.png]]
 +
 
 +
== Subnets ==
 +
 
 +
*To use a subnet (IPv4 as well as IPv6) in ESXi, at least one additional single IP is required as a router VM, since ESXi itself cannot route. When ordering a subnet, please make sure to note that it is required for ESXi and should be routed on the additional single IP.
 +
 
 +
'''IMPORTANT''' Since IPv6 subnets are routed to link-local addresses (MAC-based), it is only possible to use IPv6 in a limited way (ie. in a single VM).
 +
 
 +
=== IPv4 ===
 +
 
 +
The confirmation email of the subnet contains (for example) the following information:
 +
 
 +
<pre>
 +
Below you will find the IP subnet added to your server 192.168.13.156.
 +
 
 +
Subnet: 192.168.182.16 /28
 +
Mask: 255.255.255.240
 +
Broadcast: 192.168.182.31
 +
 
 +
Usable IP addresses:
 +
192.168.182.17 to 192.168.182.30
 +
</pre>
 +
 
 +
You do NOT get a separate MAC for each IP from a subnet.
 +
 
 +
=== IPv6 ===
 +
 
 +
All servers come with a /64 IPv6 subnet. To see which IPv6 subnet your server has please check the "IPs" tab of the server in [[Robot/en|Robot]].
 +
 
 +
If you ordered your server before February 2013 this subnet can be ordered (freely) via [[Robot/en|Robot]] and will be automatically activated.
 +
 
 +
The IPv6 subnet is routed to the default link-local address (which is derived from the MAC address) of the main IP. Via [[Robot/en|Robot]] the routing of the IPv6 subnet can be switched to the link-local address of the virtual MAC (in other words, the additional single IP). This can be done in [[Robot/en|Robot]], using the same symbol which is found next to additional single IPs to request virtual MAC addresses. The host system, so the ESXi itself, receives no IPv6 address. This is neither necessary nor possible because ESXi can not work with a fe80::1 gateway.
 +
 
 +
In order to use these IP addresses in virtual machines, a "router VM" supplemented by an additional virtual NIC from the new subnet is necessary. The subnet itself requires a new vSwitch in ESXi to which all VMs in the subnet will be connected.
 +
 
 +
=== Notes ===
 +
 
 +
The network card type for the router VM should not be VMXNET2 or VMXNET3, as otherwise the TCP performance can be very bad. As a workaround LRO in the VM can be disabled via 'disable_lro=1'. More information on this bug can be found here: [http://www.vmware.com/support/vsphere4/doc/vsp_esxi41_vc41_rel_notes.html http://www.vmware.com/support/vsphere4/doc/vsp_esxi41_vc41_rel_notes.html]
 +
 
 +
After an upgrade to VMware ESXi 5 this issue may appear again. You can address this issue by disabling Large Receive Offload (LRO) on the ESXi host:
 +
 
 +
*Log into the ESXi host with the vSphere Client.
 +
*Select the host -> Configuration -> Software:Advanced Settings.
 +
*Select Net and scroll down slightly more than half way.
 +
*Set the following parameters from 1 to 0:
 +
 
 +
Net.VmxnetSwLROSL
 +
Net.Vmxnet3SwLRO
 +
Net.Vmxnet3HwLRO
 +
Net.Vmxnet2SwLRO
 +
Net.Vmxnet2HwLRO
 +
 
 +
Reboot the ESXi host to activate these changes.
 +
 
 +
If you experience connection problems in systems with Realtek network cards then under certain circumstances deactivating offloading and activating polling can solve this. However, this also reduces the performance.
 +
 
 +
* checksum offload: deactivated
 +
* segmentation offload: deactivated
 +
* large receive offload: deactivated
 +
* device polling: enabled
 +
 
 +
=== Preparations in the vSphere client ===
 +
 
 +
1. Create a vSwitch (in the example the name "subnetz" is used)
 +
 
 +
[[File:Esxi-vswitch1.png|600px|vSwitch part 1]]<br/><br/>[[File:Esxi-vswitch2.png|vSwitch Part 2]]<br/><br/>[[File:Esxi-vswitch3.png|vSwitch Part 3]]<br/><br/>[[File:Esxi-vswitch4.png|vSwitch Part 4]]
 +
 
 +
2. Add a second NIC to the router VM. Connected network: subnetz (the previously created vSwitch)<br/><br/>[[File:Esxi-router-nic.png|600px|Esxi-router-nic.png]]
 +
 
 +
3. The NIC of the virtual machine in the subnet. Connected network: subnetz
 +
 
 +
The networking overview should show the following:
 +
 
 +
[[File:Esxi-subnet.png]]
 +
 
 +
=== Configuration of the Router VM ===
  
 
*Example of /etc/network/interfaces on the router VM
 
*Example of /etc/network/interfaces on the router VM
 +
 
  # The loopback network interface
 
  # The loopback network interface
 
  auto lo
 
  auto lo
 
  iface lo inet loopback
 
  iface lo inet loopback
 
  # The primary network interface
 
  # The primary network interface
  # WAN-NIC in VMnetwork
+
  # WAN-NIC im VMnetwork
 
  auto eth0
 
  auto eth0
 
  iface eth0 inet dhcp
 
  iface eth0 inet dhcp
 +
# for the IPv6 subnet the configuration is analog to other virtualisations
 +
iface eth0 inet6 static
 +
  address 2a01:4f8:61:20e1::2
 +
  netmask 128
 +
  gateway fe80::1
 
  # LAN NIC in Subnet
 
  # LAN NIC in Subnet
 
  auto eth1
 
  auto eth1
 
  iface eth1 inet static
 
  iface eth1 inet static
    address    192.168.182.30
+
  address    192.168.182.30
    broadcast   192.168.182.31
+
   netmask    255.255.255.240
    netmask    255.255.255.240
+
# The prefix/netmask can/must be changed according to the amount of network
 +
# segments
 +
iface eth1 inet6 static
 +
  address    2a01:4f8:61:20e1::2
 +
  netmask    64
  
 
*Example of /etc/network/interfaces of a Linux VM in the subnet
 
*Example of /etc/network/interfaces of a Linux VM in the subnet
 +
 
  # The loopback network interface
 
  # The loopback network interface
 
  auto lo
 
  auto lo
Zeile 85: Zeile 230:
 
  auto eth0
 
  auto eth0
 
  iface eth0 inet static
 
  iface eth0 inet static
    address 192.168.182.17
+
  address 192.168.182.17
    netmask 255.255.255.240
+
  netmask 255.255.255.240
    broadcast 192.168.182.31
+
  gateway 192.168.182.30
    gateway 192.168.182.30
+
iface eth0 inet6 static
 +
  address    2a01:4f8:61:20e1::4
 +
  netmask    64
 +
  gateway    2a01:4f8:61:20e1::2
  
The router VM is connected to both networks (public Hetzner and assigned subnet) and must be used as a gateway for virtual machines in the subnet. Don't forget to activate IP forwarding in the kernel:
+
 
 +
The router VM is now connected to both networks and can be used as a gateway for virtual machines in the subnet. Don't forget to activate IP forwarding in the kernel:
  
 
  echo 1 > /proc/sys/net/ipv4/ip_forward
 
  echo 1 > /proc/sys/net/ipv4/ip_forward
 +
echo 1 > /proc/sys/net/ipv6/conf/all/forwarding
  
 
To make this persistent across reboots, add the following line to /etc/sysctl.conf
 
To make this persistent across reboots, add the following line to /etc/sysctl.conf
  
 
  net.ipv4.ip_forward=1
 
  net.ipv4.ip_forward=1
 +
net.ipv6.conf.all.forwarding=1
  
The virtual machines should now be accessible from the Internet over the assigned IP.
+
The virtual machines should now be accessible (via SSH for example) via their assigned IPs.
<br>
+
  
= Installation guide =
+
= Installation Guide =
  
* When ordering your desired EQ server an optional RAID controller can be added if desired.
+
*Choose the "Rescue System" as the OS for the server you order.
  
* If you have stated in your order that you would like to install VMware ESXI you will receive an email from Hetzner when your server is ready for installation. The email will contain access details for a remote console (LARA) for the installation. A DVD drive with the Install-CD will be connected as well.
+
*If you would like RAID you can add a 4-Port RAID controller since ESXi doesn't support software RAID.
  
* Login to the LARA console and finish the installation (approx. 5 minutes)
+
*Once the server is online (you will get an email from us informing you of this) you can order a KVM Console and use that to virtually mount an ISO file of the version of ESXi you want to install. Further information on ordering a KVM Console and using it to install an OS can be found on the [[KVM-Console/en|KVM Console]] page.
  
[[Datei:Esxi-installed.png]]
+
Afterwards, the following screen should be displayed:
  
* After the reboot, set a password using the LARA console (mind the disk order!). This is the root password for SSH (if you enable it) as well as the password for the vSphere Client.
+
[[File:Esxi-installed.png|500px|ESXi installed]]
  
* After successful installation you can order up to three additional IP addresses (with a proper RIPE reason). If you also include that you need specials MAC addresses for ESXi, they may be assigned as well. For more than four IP adresses you need a subnet.
+
*After a reboot login using the password that you entered during the installation. This is the root password for SSH as well as the password for the VMware vSphere Client (requires Windows). This can then be downloaded separately via a browser.
  
* Configure the MAC addresses in your virtual servers with the correct IP addresses through vSphere. If you do this right, even DHCP works!
+
[[File:Esxi-vsphere.png]]
 +
 
 +
*After successful installation you can order up to three additional single IP addresses via Robot. You can get a MAC address for your additional single IP through Robot under the "IPs" tab. You will see a small button next to the IP address. Clicking on that button will give the IP address a virtual MAC.
 +
*Configure the MAC addresses of the virtual servers with the corresponding IP addresses through vSphere. Once this is done even DHCP will work via the Hetzner network!
  
 
For additional information regarding ESXi and its usage, please refer to the [http://www.vmware.com/products/esxi/ official website]
 
For additional information regarding ESXi and its usage, please refer to the [http://www.vmware.com/products/esxi/ official website]
Zeile 121: Zeile 274:
 
== Manual installation of updates ==
 
== Manual installation of updates ==
  
The normal way to install updates is via the Host Update Utility. Because uploading an update of hundreds of megabytes with a DSL connection takes very long, the following is a guide on how to manually install updates on the system. This guide comes with absolutely no warranty for correctness or usability! Use it at your own risk.
+
The installation of updates in the free version can only be done via the console or via VMware Go. An update can be several hundred megabytes, which can take a long time with a standard DSL connection, which is why the following guide can help. This is done at your own risk. There is no warranty or guarantee for correctness!
 +
 
 +
Prerequisite is SSH activated access and that the system is in maintenance mode. This can be activated via:
 +
 
 +
vim-cmd hostsvc/maintenance_mode_enter
 +
 
 +
=== Updating vSphere 5.0 to 5.1 ===
 +
 
 +
First of all the VMware-ESXi-5.1.0-799733-depot.zip update needs to be downloaded from the [http://www.vmware.com/patchmgr/download.portal VMWare Updates Page] and saved on the ESXi Host.
 +
 
 +
Once all the VMs have been shut down and the system has entered the maintenance mode via "vim-cmd hostsvc/maintenance_mode_enter", the update can be installed in two ways. The following command refreshes the system and removes all the packages that are not included in the update. This is equivalent to a new installation.
 +
 
 +
esxcli software profile install -d /vmfs/volumes/datastore1/VMware-ESXi-5.1.0-799733-depot.zip -p ESXi-5.1.0-799733-standard
 +
 
 +
Alternatively, only the packages containted in the update can be updated to their new versions, leaving all other packages intact.
 +
 
 +
esxcli software profile update -d /vmfs/volumes/datastore1/VMware-ESXi-5.1.0-799733-depot.zip -p ESXi-5.1.0-799733-standard
 +
 
 +
Finally, the system must be rebooted. When the VMs are turned on for the first time after the reboot, it is possible that a message appears noting that the VM was copied or moved. This happens because the UUIDs get changed during the update. You can safely select "VM was moved" in this situation. See also: [http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1010675 VMware Help Page]
 +
 
 +
[[Kategorie:VMWare ESXi]]
 +
 
 +
=== Installation of Patches ===
 +
 
 +
After the patches have been transferred onto the system, they can be installed. It is important that the full path name is entered, eg.:
 +
 
 +
esxcli software vib install --depot="/vmfs/volumes/datastore1/patches/ESXi510-201210001.zip"
 +
Installation Result
 +
Message: The update completed successfully, but the system needs to be rebooted for the changes to be effective.
 +
Reboot Required: true
 +
[...]
 +
 
 +
After a reboot the maintenance mode needs to be exited:
 +
 
 +
vim-cmd hostsvc/maintenance_mode_exit
 +
 
 +
== Monitoring RAID controller ==
 +
 
 +
=== 3ware Controller ===
 +
 
 +
For the 3ware controller there exists both a CIM Provider as well as a CLI. The 64-bit CLI for Linux can be used from version 9.5.2 onwards.
 +
 
 +
''Note'': 3ware controllers are only supported by ESXi 5.0 via an external driver.
 +
 
 +
=== Adaptec Controller ===
 +
 
 +
For Adaptec controllers the CIM Provider and the CLI (arcconf) must be manually installed. Required is an up-to-date version of the driver. An installation guide can be found on the [http://download.adaptec.com/pdfs/installation_guides/vmware_esxi_41_cim_remotearcconf_installation_guide_3_2011.pdf Adaptec website] (PDF)
 +
 
 +
*RAID driver [https://my.vmware.com/web/vmware/details?downloadGroup=DT-ESXi50-PMC-Sierra-aacraid-52129800&productId=229 Version 5.2.1.29800]
 +
*[http://www.adaptec.com/en-us/speed/raid/storage_manager/asm_linux_x64_v7_30_18837_tgz.htm Remote Arcconf]
 +
*[http://www.adaptec.com/en-us/speed/raid/storage_manager/cim_vmware_v7_31_18856_zip.htm Adaptec CIM Providers]
 +
 
 +
Monitoring can be achieved through the installation of remote ARCCONF via a Windows/Linux system.
 +
 
 +
$ arcconf GETCONFIG 1 AD
 +
 
 +
=== LSI Controller ===
 +
 
 +
LSI provides a so-called CIM/SMIS provider. After the installation the hardware monitoring page in the vSphere client displays the status of the RAID. An active alarm is however, only possible in the paid version and when running vCenter.
 +
 
 +
Alternatively, the command line tool '''MegaCLI''' can be installed, which is also used to manage the RAID controller. A script can be used to automate the displaying of status information. This script and notifications must be run from another server.
 +
 
 +
== Parallel operation of onboard controller/hardware RAID ==
 +
 
 +
During the installation ESXi only "sees" one type of storage, so either the onboard SATA controller or an additional RAID controller. If drives are connected to both then the hardware controller is prioritized and the drives connected to the onboard controller are invisible. By manually loading the appropriate kernel module these drives can still be used.
  
Prerequisite is SSH activated access and wget.
+
/sbin/vmkload_mod ahci
  
First you need to download the patch from [http://support.vmware.com/selfsupport/download/ VMware self support site]. Choose ESXi and search. The major upgrade from 4.0 to 4.1 can be found on the regular [http://www.vmware.com/go/esxi ESXi download site] or already repacked to tar.gz from [http://download.hetzner.de/tools/vmware/upgrade-from-ESXi4.0-to-4.1.0-0.0.260247-release.tar.gz download.hetzner.de] (no warranty, use at your own risk)
+
To have this module loaded automatically during start, the line above must be added to /etc/rc.local and /sbin/auto-backup.sh.
  
Since the files are compressed with zip, which does not exist on ESXi, the updates have to be repacked with tar. So using either a Linux vm or another server unzip the downloaded files and compress them again using tar and gzip. The tar.gz is then transferred to the ESXi host using either wget (requires webserver) or rsync.
+
= Hardware change =
  
The archive can then be extracted in a directory of your choice:
+
== Change MAC address ==
  
/vmfs/volumes/4a42186d-2d171bfc-6d01-001b213b649f/updates/ESXi400-201003001 # ls -all
+
In the event of a hardware (ex)change, especially the motherboard, it should be noted that the ESXi host retains its original MAC address. This leads to problems as the switch will not automatically forward the correct new main IP to the server, as the MAC address that is being broadcast is incorrect. The MAC address needs to be reset via the ESXi shell. There are several approches to do this, listed by the following [http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1031111 VMWare Knowledge Base Article]. The most elegant solution is when the ESXi host automatically recognizes the new MAC address when changing platforms and uses that. The following command can be used for that:
drwxr-xr-x    1 root    root                560 Apr  6 18:29 .
+
drwxr-xr-x    1 root    root                560 Apr  6 18:27 ..
+
drwxr-xr-x    1 root    root                420 Apr  6 18:29 embeddedEsx
+
-rw-r--r--    1 root    root              4946 Mar 24 15:25 metadata.zip
+
  
Example ESXi 4.1 Upgrade:
+
esxcfg-advcfg -s 1 /Net/FollowHardwareMac
  
/vmfs/volumes/4c4559de-9e2a8780-a90b-001b215f7a57/upgrade-40-to-41 # ls -l
+
Either perform this command before the platform change or, if the change has already happened, there are two options:
-rw-r--r--    1 1000    1000              4934 May 19 00:30 metadata.zip
+
-r--r--r--    1 1000    1000              1507 May 19 00:29 upgrade_info.xml
+
-rw-r--r--    1 1000    1000            1351442 May 19 00:29 vmware- esx-esxupdate-esxi-4.1.0-0.0.260247.i386.vib
+
-rw-r--r--    1 1000    1000          89605022 May 19 00:29 vmware-esx-firmware-4.1.0-0.0.260247.i386.vib
+
-rw-r--r--    1 1000    1000          120626062 May 19 00:30 vmware-esx-tools-light-4.1.0-0.0.260247.i386.vib
+
  
After all guests have been shutdown, the maintenance mode must be activated
+
* Order a KVM Console console and enable the ESXi shell and then press Alt + F1 to switch to the console and enter the command. Afterwards pressing Alt + F2 will bring you back to the GUI.
 +
* Temporarily teach the switch the new MAC address by booting into the Rescue System and then back into the ESXi host. As a result, the ESXi host is now reachable again via the main IP, but only for a limited amount of time. The length of this time depends on how long the switch takes to delete the ARP cache entry for this MAC address. Normally there is enough time to login via SSH and execute the command, assuming SSH access has been enabled. However, even this is configurable, as connecting via the ESXi client would be possible again.
  
../updates/ESXi400-201003001 # vim-cmd hostsvc/maintenance_mode_enter
+
With either option a restart is required afterwards. This can be initiated via the console:
'vim.Task:haTask-ha-host-vim.HostSystem.enterMaintenanceMode-1852795269'
+
  
Now the update can be installed
+
reboot
  
../updates/ESXi400-201003001 # esxupdate update -m metadata.zip
+
After a restart the MAC address should be set correctly and this can be verified in the ESXi shell via the following command:
metadata.zip                                                  ############### [100%]
+
vmware-esx-firmware-4.0.0-1.12.244038.i386.vib                ############### [100%]
+
Unpacking deb_vmware-esx-firmware_4.0.0-1.12.244038.vib        ############### [100%]
+
Installing packages :deb_vmware-esx-firmware_4.0.0-1.12.244038 ############### [100%]
+
The update completed successfully, but the system needs to be rebooted for the changes to be effective.
+
/vmfs/volumes/4a42186d-2d171bfc-6d01-001b213b649f/updates/ESXi400-201003001 #
+
  
 +
esxcfg-vmknic -l
  
After a reboot, the update is complete. This reduces the downtime of your ESXi to about 5 minutes.
+
The new MAC address should show up next to the main IP.
  
This guide was first posted by Tischi in the [http://forum.hetzner.de/wbb2/thread.php?threadid=14372 Hetzner Forum]
+
[[Kategorie:Dedizierte Server]]
 +
[[Kategorie:Betriebssysteme]]
 +
[[Kategorie:Virtualisierung]]

Version vom 3. Februar 2020, 18:55 Uhr

Inhaltsverzeichnis

General information

Hardware

  • VMware vSphere Hypervisor 5.x and 6.0 (ESXi) use a filesystem called vmfs5 (formerly vmfs3), which uses GPT and supports drive sizes of up to 64 TiB (formerly 2 TiB).
  • Software RAID is *not* supported by ESXi.
  • The free version of ESXi starting with 5.5 supports a maximum of 4 TB RAM (formerly 32 GB RAM).

Network configuration

  • VMware vSphere Hypervisor is an "operating system" designed for pure virtualization and supports neither NAT nor routing. Therefore, only a true bridged setup can be used.
  • To use a subnet an additional single IP needs to be setup as a router VM.

Installation

  • The installation and configuration of ESXi takes about 20-30 minutes, even without any prior knowledge. Please check out the Installation Guide for further information on installing ESXi.

License

  • After installation the system has a test license, which expires after 60 days. For the long-term usage of the OS a separate (free) license is required. This can be requested by registering on the VMware website:
  • The license can be entered via the vSphere Client, under the tab "Configuration", the section "Software" and then under the menu item "Licensed functions". After selecting this you can click on the "Edit" button to the right and then enter the license key. Entering license with vSphere Client

Hetzner specific information

Installed hardware

The Dell PowerEdge models DX150, DX151, DX290, DX291  and DX141 are certified and compatible with all versions from 5.0 onwards. DX152 and DX292 are compatible from 6.5 onwards.

All other models are not certified by VMware, yet in most cases they can be run with VMware vSphere/ESXi.

Compatibility

(All information is subject to change)

Model vSphere/ESXi Version
AX50-SSD/AX60-SSD from 6.5a additional Intel NIC required
AX41/AX51
from 6.7 Update 2
(potentially an additional NIC required)
AX100 from 6.7
AX160 from 6.5 Update 2 / 6.7
PX92 from 6.5
PX91/PX121 from 5.5 Update 1
PX90/PX120 from 5.1
PX61 from 6.0 Update 2
PX62
from 6.5 Update 2 / 6.7 Update 1
PX60/PX70 from 5.5 Update 1
EX61
from 6.5
EX52/EX62
from 6.5 Update 2 / 6.7 Update 1
EX41/EX41S/EX51/EX42
from 5.5 Update 3 / 6.0 Update 2
(potentially an additional NIC required)
EX40/EX60 5.0 - 5.1 Update 2, since 5.5 additional NIC required
EX6/EX6S/EX8/EX8S from 5.0
EX4/EX4S/EX10 from 5.0 (with additional NIC)
EQ4/EQ6/EQ8/EQ9 5.0 - 5.1 Update 2, since 5.5 additional NIC required

NOTES:

  • The models mentioned above should work with the appropriate version. However, newer versions are not necessarily compatible. To make sure please gather your own information.
  • The Realtek driver present in ESXi 5.0 to 5.1 Update 2 was just a Tech Demo that has not been updated. Therefore, for smooth operation a server with an Intel NIC (PX60, PX91, DX151 etc.) is recommended.
  • The installation of vSphere 5.5 or newer on the EX40, EX60 and some EX41/EX51 models requires a custom installation ISO that includes drivers for the Realtek NIC (http://www.bussink.ch/?p=1228). Alternatively, an additional compatible NIC can be installed. Please find the cost of an additional compatible NIC here: Root Server Hardware
  • Installing ESXi on the older DS or X servers is not possible.
  • When installing on a server with multiple identical drives, ESXi can potentially show the drives in a different order than the BIOS does. Should the screen be blank with only a blinking cursor after the installation is done and you restart, try booting from the other drive.

Network Configuration

  • For connectivity between multiple ESXi servers within the same subnet, host routes via gateway are necessary due to network security restrictions.
host A
esxcfg-route -a <IP Host B> 255.255.255.255 <Gateway IP>
host B
esxcfg-route -a <IP Host A> 255.255.255.255 <Gateway IP>

Single IP addresses

  • IP addresses are, by default, statically mapped to the MAC address of the host. It is possible however to get separate MAC addresses for the additional single IPs via Robot. These can then be configured for the virtual machines. To get these MAC addresses assigned, log in to Robot, choose "Server" from the menu on the left, select the desired server, and then click on the "IPs" tab. Here you can click on the icon next to the single IP address to get a MAC address.

Esxi-mac-setzen.png

Subnets

  • To use a subnet (IPv4 as well as IPv6) in ESXi, at least one additional single IP is required as a router VM, since ESXi itself cannot route. When ordering a subnet, please make sure to note that it is required for ESXi and should be routed on the additional single IP.

IMPORTANT Since IPv6 subnets are routed to link-local addresses (MAC-based), it is only possible to use IPv6 in a limited way (ie. in a single VM).

IPv4

The confirmation email of the subnet contains (for example) the following information:

Below you will find the IP subnet added to your server 192.168.13.156.

Subnet: 192.168.182.16 /28
Mask: 255.255.255.240
Broadcast: 192.168.182.31

Usable IP addresses:
192.168.182.17 to 192.168.182.30

You do NOT get a separate MAC for each IP from a subnet.

IPv6

All servers come with a /64 IPv6 subnet. To see which IPv6 subnet your server has please check the "IPs" tab of the server in Robot.

If you ordered your server before February 2013 this subnet can be ordered (freely) via Robot and will be automatically activated.

The IPv6 subnet is routed to the default link-local address (which is derived from the MAC address) of the main IP. Via Robot the routing of the IPv6 subnet can be switched to the link-local address of the virtual MAC (in other words, the additional single IP). This can be done in Robot, using the same symbol which is found next to additional single IPs to request virtual MAC addresses. The host system, so the ESXi itself, receives no IPv6 address. This is neither necessary nor possible because ESXi can not work with a fe80::1 gateway.

In order to use these IP addresses in virtual machines, a "router VM" supplemented by an additional virtual NIC from the new subnet is necessary. The subnet itself requires a new vSwitch in ESXi to which all VMs in the subnet will be connected.

Notes

The network card type for the router VM should not be VMXNET2 or VMXNET3, as otherwise the TCP performance can be very bad. As a workaround LRO in the VM can be disabled via 'disable_lro=1'. More information on this bug can be found here: http://www.vmware.com/support/vsphere4/doc/vsp_esxi41_vc41_rel_notes.html

After an upgrade to VMware ESXi 5 this issue may appear again. You can address this issue by disabling Large Receive Offload (LRO) on the ESXi host:

  • Log into the ESXi host with the vSphere Client.
  • Select the host -> Configuration -> Software:Advanced Settings.
  • Select Net and scroll down slightly more than half way.
  • Set the following parameters from 1 to 0:
Net.VmxnetSwLROSL
Net.Vmxnet3SwLRO
Net.Vmxnet3HwLRO
Net.Vmxnet2SwLRO
Net.Vmxnet2HwLRO

Reboot the ESXi host to activate these changes.

If you experience connection problems in systems with Realtek network cards then under certain circumstances deactivating offloading and activating polling can solve this. However, this also reduces the performance.

  • checksum offload: deactivated
  • segmentation offload: deactivated
  • large receive offload: deactivated
  • device polling: enabled

Preparations in the vSphere client

1. Create a vSwitch (in the example the name "subnetz" is used)

vSwitch part 1

vSwitch Part 2

vSwitch Part 3

vSwitch Part 4

2. Add a second NIC to the router VM. Connected network: subnetz (the previously created vSwitch)

Esxi-router-nic.png

3. The NIC of the virtual machine in the subnet. Connected network: subnetz

The networking overview should show the following:

Esxi-subnet.png

Configuration of the Router VM

  • Example of /etc/network/interfaces on the router VM
# The loopback network interface
auto lo
iface lo inet loopback
# The primary network interface
# WAN-NIC im VMnetwork
auto eth0
iface eth0 inet dhcp
# for the IPv6 subnet the configuration is analog to other virtualisations
iface eth0 inet6 static
 address 2a01:4f8:61:20e1::2
 netmask 128
 gateway fe80::1
# LAN NIC in Subnet
auto eth1
iface eth1 inet static
 address     192.168.182.30
 netmask     255.255.255.240
# The prefix/netmask can/must be changed according to the amount of network
# segments
iface eth1 inet6 static
 address    2a01:4f8:61:20e1::2
 netmask    64
  • Example of /etc/network/interfaces of a Linux VM in the subnet
# The loopback network interface
auto lo
iface lo inet loopback
# The primary network interface
auto eth0
iface eth0 inet static
 address 192.168.182.17
 netmask 255.255.255.240
 gateway 192.168.182.30
iface eth0 inet6 static
 address    2a01:4f8:61:20e1::4
 netmask    64
 gateway    2a01:4f8:61:20e1::2


The router VM is now connected to both networks and can be used as a gateway for virtual machines in the subnet. Don't forget to activate IP forwarding in the kernel:

echo 1 > /proc/sys/net/ipv4/ip_forward
echo 1 > /proc/sys/net/ipv6/conf/all/forwarding

To make this persistent across reboots, add the following line to /etc/sysctl.conf

net.ipv4.ip_forward=1
net.ipv6.conf.all.forwarding=1

The virtual machines should now be accessible (via SSH for example) via their assigned IPs.

Installation Guide

  • Choose the "Rescue System" as the OS for the server you order.
  • If you would like RAID you can add a 4-Port RAID controller since ESXi doesn't support software RAID.
  • Once the server is online (you will get an email from us informing you of this) you can order a KVM Console and use that to virtually mount an ISO file of the version of ESXi you want to install. Further information on ordering a KVM Console and using it to install an OS can be found on the KVM Console page.

Afterwards, the following screen should be displayed:

ESXi installed

  • After a reboot login using the password that you entered during the installation. This is the root password for SSH as well as the password for the VMware vSphere Client (requires Windows). This can then be downloaded separately via a browser.

Esxi-vsphere.png

  • After successful installation you can order up to three additional single IP addresses via Robot. You can get a MAC address for your additional single IP through Robot under the "IPs" tab. You will see a small button next to the IP address. Clicking on that button will give the IP address a virtual MAC.
  • Configure the MAC addresses of the virtual servers with the corresponding IP addresses through vSphere. Once this is done even DHCP will work via the Hetzner network!

For additional information regarding ESXi and its usage, please refer to the official website

Manual installation of updates

The installation of updates in the free version can only be done via the console or via VMware Go. An update can be several hundred megabytes, which can take a long time with a standard DSL connection, which is why the following guide can help. This is done at your own risk. There is no warranty or guarantee for correctness!

Prerequisite is SSH activated access and that the system is in maintenance mode. This can be activated via:

vim-cmd hostsvc/maintenance_mode_enter

Updating vSphere 5.0 to 5.1

First of all the VMware-ESXi-5.1.0-799733-depot.zip update needs to be downloaded from the VMWare Updates Page and saved on the ESXi Host.

Once all the VMs have been shut down and the system has entered the maintenance mode via "vim-cmd hostsvc/maintenance_mode_enter", the update can be installed in two ways. The following command refreshes the system and removes all the packages that are not included in the update. This is equivalent to a new installation.

esxcli software profile install -d /vmfs/volumes/datastore1/VMware-ESXi-5.1.0-799733-depot.zip -p ESXi-5.1.0-799733-standard

Alternatively, only the packages containted in the update can be updated to their new versions, leaving all other packages intact.

esxcli software profile update -d /vmfs/volumes/datastore1/VMware-ESXi-5.1.0-799733-depot.zip -p ESXi-5.1.0-799733-standard

Finally, the system must be rebooted. When the VMs are turned on for the first time after the reboot, it is possible that a message appears noting that the VM was copied or moved. This happens because the UUIDs get changed during the update. You can safely select "VM was moved" in this situation. See also: VMware Help Page

Installation of Patches

After the patches have been transferred onto the system, they can be installed. It is important that the full path name is entered, eg.:

esxcli software vib install --depot="/vmfs/volumes/datastore1/patches/ESXi510-201210001.zip"
Installation Result
Message: The update completed successfully, but the system needs to be rebooted for the changes to be effective.
Reboot Required: true
[...]

After a reboot the maintenance mode needs to be exited:

vim-cmd hostsvc/maintenance_mode_exit

Monitoring RAID controller

3ware Controller

For the 3ware controller there exists both a CIM Provider as well as a CLI. The 64-bit CLI for Linux can be used from version 9.5.2 onwards.

Note: 3ware controllers are only supported by ESXi 5.0 via an external driver.

Adaptec Controller

For Adaptec controllers the CIM Provider and the CLI (arcconf) must be manually installed. Required is an up-to-date version of the driver. An installation guide can be found on the Adaptec website (PDF)

Monitoring can be achieved through the installation of remote ARCCONF via a Windows/Linux system.

$ arcconf GETCONFIG 1 AD

LSI Controller

LSI provides a so-called CIM/SMIS provider. After the installation the hardware monitoring page in the vSphere client displays the status of the RAID. An active alarm is however, only possible in the paid version and when running vCenter.

Alternatively, the command line tool MegaCLI can be installed, which is also used to manage the RAID controller. A script can be used to automate the displaying of status information. This script and notifications must be run from another server.

Parallel operation of onboard controller/hardware RAID

During the installation ESXi only "sees" one type of storage, so either the onboard SATA controller or an additional RAID controller. If drives are connected to both then the hardware controller is prioritized and the drives connected to the onboard controller are invisible. By manually loading the appropriate kernel module these drives can still be used.

/sbin/vmkload_mod ahci

To have this module loaded automatically during start, the line above must be added to /etc/rc.local and /sbin/auto-backup.sh.

Hardware change

Change MAC address

In the event of a hardware (ex)change, especially the motherboard, it should be noted that the ESXi host retains its original MAC address. This leads to problems as the switch will not automatically forward the correct new main IP to the server, as the MAC address that is being broadcast is incorrect. The MAC address needs to be reset via the ESXi shell. There are several approches to do this, listed by the following VMWare Knowledge Base Article. The most elegant solution is when the ESXi host automatically recognizes the new MAC address when changing platforms and uses that. The following command can be used for that:

esxcfg-advcfg -s 1 /Net/FollowHardwareMac

Either perform this command before the platform change or, if the change has already happened, there are two options:

  • Order a KVM Console console and enable the ESXi shell and then press Alt + F1 to switch to the console and enter the command. Afterwards pressing Alt + F2 will bring you back to the GUI.
  • Temporarily teach the switch the new MAC address by booting into the Rescue System and then back into the ESXi host. As a result, the ESXi host is now reachable again via the main IP, but only for a limited amount of time. The length of this time depends on how long the switch takes to delete the ARP cache entry for this MAC address. Normally there is enough time to login via SSH and execute the command, assuming SSH access has been enabled. However, even this is configurable, as connecting via the ESXi client would be possible again.

With either option a restart is required afterwards. This can be initiated via the console:

reboot

After a restart the MAC address should be set correctly and this can be verified in the ESXi shell via the following command:

esxcfg-vmknic -l

The new MAC address should show up next to the main IP.



© 2020. Hetzner Online GmbH. Alle Rechte vorbehalten.