Sunday, 13 April 2014

Adding and mounting a new disk in Linux to expand storage and house the contents of /home

So you want to add another disk to your Linux install. There could be several reasons for this, for instance running out of space on your existing setup, or wanting to split your system across multiple drives.

In my case I'd setup a new cPanel instance and wanted more capacity for the content in /home using a second disk. As such this post also explains how to move the existing /home content onto the newly added disk, such that everything continues to work like it did before.

I'll start by assuming you've added the new drive to your server, whether physically or as a new virtual disk, in your chosen hyper-visor.

Check that the system has detected the new disk by running :

    ls /dev/sd*

You'll see a list of all the physical disks detected, and the partitions on them. The first drive is /dev/sda, and if that has two partitions on it currently you'll see /dev/sda1 and /dev/sda2 listed. This increments for each drive, so a newly added second drive will be /dev/sdb.

You can check and ensure /dev/sdb has nothing on it already by running :

    fdisk -l

which will display the disk and partition details, and indicate that /dev/sdb doesn't contain a valid partition table. So lets get a partition added. Run :

    fdisk /dev/sdb

then at the prompt press n for New. You'll be asked what type of partition you want, so press p for primary. Now you'll be asked to select how much of the disk to use, these default to the first and last cylinders, so if you're using the entire disk you can simply press return on both and use the defaults. Finally at the last prompt press w to write the table to disk and exit. Note, if you quit without pressing w then nothing will be changed.

Now if you check in /dev you'll see a newly added /dev/sdb1 listed, but before we can use it we need to format it. To format it as ext3 run :

    /sbin/mkfs.ext3 -L /home /dev/sdb1

which will format the disk. The "-L /home" part of the command specifies the volume label, so if you're not using this disk for your /home folder then you might want to adjust it accordingly.

Now we want to move the existing contents of /home onto the newly formatted disk, but obviously new and old can't be /home at the same time so...

    mkdir /mnt/home
    mount /dev/sdb1 /mnt/home

to create a temporary mount point for the new drive and then mount it. Now copy the data with :

    cp -Ra /home/* /mnt/home

and then check in /mnt/home that the data is all there. Now we're finished with the temporary mount point so we can remove it with :

    umount /mnt/home
    rmdir /mnt/home

Before we can mount the new disk to its proper home we need to get rid of the existing /home folder, so let's rename it rather than delete it :

    mv /home /home-backup

which also means that if things go wrong you can switch back to the original data. Then, because we still need a folder to mount the new drive to :

    mkdir /home

To permanently mount the new disk we need to edit the fstab file, otherwise if we simply mount it then it won't reload next time the server is booted. Edit /etc/fstab and then add the following line :

    /dev/sdb1    /home    ext3    defaults    0 2

This then mounts the drive partition /dev/sdb1 to the folder /home using the ext3 file system and using the default options. The 0 relates to the archiving frequency (generally set to 0) and 2 controls the order that fsck checks for errors on the drive at boot time. You'd set that to 1 for the boot device (so not relevant here), or 2 for a subsequent drive. Many people blogging about this set it to 0 which disables checking, which I wouldn't recommend for anything other than a drive you were using short term.

Once that's saved we just need to test that it's working. You could reboot, but personally I prefer to run :

    mount -a

which will execute all of the entries listed in fstab (but do nothing where something's already mounted). If your new entry is OK then it should mount the new disk and make the contents accessible via /home, or if there's a problem then it will indicate where the issue is so you can edit /etc/fstab again to correct the issue.

So now you have a newly mounted drive, containing the same folder structure as before, but with the /home content stored on a separate drive. If you run :


you'll see the newly added mount point and drive, and the disk usage / availability.

The above steps were tested on a CentOS 6.4 Hyper-V virtual machine, but should obviously work on most Linux installs.

Sunday, 30 March 2014

Migrating CentOS / cPanel from VMWare to Hyper-V

To migrate a CentOS / cPanel virtual machine from VMWare to Hyper-V there are a few steps you need to remember if you want to keep things easy.

The following steps assume you're attempting to migrate an existing CentOS installation, between host servers in the same network, eg you don't need or want to change any settings on the guest server like IP addresses etc. Simply convert, copy, import and power up.

Before shutting down the guest, it's advisable to uninstall the VMWare Tools from the server if they have been installed. While not always an issue, there are some reports of people having difficulties uninstalling it once the server has been migrated. Additionally, I'd suggest installing the Hyper-V Linux Integration Services prior to migration (current version at time of writing is 3.5 available here This ensures when you power up the guest in Hyper-V that you don't have any issues with it recognising the network adaptor.

Linux installations hard code the MAC address of the NIC alongside the IP address details, which means if the MAC address changes (as it will by default when moving to another host), you'll find your network settings no longer work. Make a note of the current MAC address, either from the network settings in VMWare, or by looking in /etc/sysconfig/networking/devices/ifcfg-eth0.

Shutdown the server being migrated, and copy the files to the new Hyper-V server to convert them. You could convert them on the old server, but it's probably older and slower (hence you migrating) so moving them now saves time.

To convert the files I recommend using WinImage. Simply point it at the target vmdk file, tell it where to put the resulting .vhd file and its name, and it gets on with it. Note, some people report issues converting single large vmdk files (I don't know if this applies to WinImage), in which case the advise is to use VMWare Converter to split them into smaller files before the conversion. Once finished, copy the now converted .vhd file into the folder you'll create as part of the setup for the VM in the next step.

While the conversion is taking place create the new Hyper-V VM. At time of writing it will need to be a generation 1 virtual machine if you're running 2012 R2, as CentOS doesn't support generation 2 machines. Create the VM as normal, but when prompted to configure the HDD choose the option to Attach a virtual hard disk later.

Once configured, go into the Settings for the new VM. Newer versions of CentOS may be fine, but for 5.10 I found the NIC must be configured as a Legacy Network Adaptor, so remove the Network Adaptor that's there and then add a Legacy Network Adaptor instead. Within the Adaptor settings, go into the Advanced Features and statically assign the MAC address to the one you previously recorded from VMWare. Finally, select IDE Controller 0, select Hard Drive and click Add, then click Browse and find the .vhd file you converted earlier. Obviously if you have multiple drives then repeat this for each of them.

Once all this is done you can start the VM. You should find it starts successfully and works exactly the same as previously without the need to adjust anything on the server.

Tuesday, 27 August 2013

Querying DNS RecordData properties in PowerShell

Using the Get-DnsServerResourceRecord cmdlet it's simple to retrieve the records in a domain, and by combining it with Where-Object it's simple to filter by most of the properties of the zone as well. For instance :

    Get-DnsServerResourceRecord -ZoneName "" | Where-Object {$_.RecordType -eq "MX"}

will give you all the MX records within the zone. We can equally filter by HostName, DistinguishedName and a few others since they're simple string values.

What happens when we want to query a zone on RecordData? For example, how would we find all the records that point to "", or MX records that are using "". This data isn't stored as a normal string, so it can't be queried in the same way. Using Get-Member to find the properties of Get-DnsServerResourceRecord you'll find that the RecordData property type is "CimInstance#Instance", not string like others mentioned above.

What this essentially means is that the RecordData property has properties within it, and it's these that we need query. Each type of record has its own individual properties corresponding to the type of data held within it.

Depending on the type of record you're querying there are different property names to use. To find the list use :

    $records = Get-DnsServerResourceRecord -ZoneName ""
    $records.RecordData | Get-Member

You'll see some have one relevant property, for instance IPv4Address, while others which hold more info have multiple, for instance MailExchange and Preference for MX records.

Note, you'll only see those values that exist within the zone you used above. So if the zone doesn't have an MX record you won't see any MX record details listed.

To query using them you add the property after RecordData, for instance :

    Get-DnsServerResourceRecord -ZoneName "" | Where-Object {$_.RecordData.IPv4Address -eq ""}


    Get-DnsServerResourceRecord -ZoneName "" | Where-Object {$_.RecordData.MailExchange -match ""}

You see it's fairly straight forward to query these, it's just a question of finding them in the first place. Below is a list of the most common record types, it's not every single one possible but it should cover most situations.

A Record            IPv4Address
AAAA Record      IPv6Address
MX Record          MailExchange, Preference
CNAME Record    HostNameAlias
SRV Record        DomainName, Port, Priority, Weight
TXT Record        DescriptiveText
SOA Record       PrimaryServer, ExpireLimit, MinimumTimeToLive,   

                        RefreshInterval, ResponsiblePerson, RetryDelay, Serial Number
NS Record         NameServer
PTR Record        PtrDomainName

If you require any additional record types simply use Get-Member as listed above on a zone containing the required properties.


Thursday, 18 July 2013

Removing a secondary zone from all DNS servers in an AD domain with PowerShell 3.0

Following on from the last blog "Adding secondary zones to all DNS servers in an AD domain with PowerShell 3.0" I'll move on to removing secondary zones from all the DNS servers in an AD domain.

Much of the code is very similar to that used when creating a new secondary, so I won't bother repeating those bits. As before you obviously need to retrieve the current list of DNS servers on the network and then work through the list.

To delete the zone itself we use the command :

    Remove-DnsServerZone -Name $domain -ComputerName $dnsserver -Force

but to add a little complication I also wanted to log the currently configured master server for the zone before deleting it. With that logged if we accidentally delete a zone it's easy to find where it pointed previously and set it up again.

Unfortunately as far as I can find there's currently no way to retrieve this info using PowerShell, so I had to resort to using the old friend of DNS scripting, dnscmd :

    $master=(dnscmd /zoneinfo $domain) -split '[,]' | ? {$_ -like '*addr=*'}
    write-output "Current master server for $domain is $master" | out-file $logfile -append

This retrieves the zoneinfo data for the domain being deleted, grabs the line containing "addr=" which lists the master servers, and then outputs that information to a log file.

You can download the completed script, which includes logging and error trapping, from

Monday, 1 July 2013

Adding secondary zones to all DNS servers in an AD domain with PowerShell 3.0

One of the advantages of AD Integrated DNS is that adding or editing a zone on one server automatically replicates that action onto all the others, but that only works when the domain is an AD Integrated Primary. If you're dealing with secondary zones that doesn't apply, and things work like they do in a non-AD environment.

Adding a secondary zone needs to be done on each DNS server within the AD domain individually, which if you have a lot of DNS servers could be a lot of work. Add to that the more servers you're working with, the more chance of a mistake being made on one of them.

To get around this I wanted to script a way to create a secondary zone on all the DNS servers at once.

The first step is getting a list of the DNS servers on the domain. You could manually create and maintain a list of servers, but I prefer to assume things will change and get the script to allow for this. To generate a list of DNS servers I use the following :

    $dnserversldap = [ADSI]"LDAP://ou=Domain Controllers,dc=mydomain,dc=co,dc=uk"
    $objsearcher = new-object system.directoryservices.directorysearcher
    $objsearcher.searchroot = $dnserversldap
    $objsearcher.filter = "(objectcategory=computer)"
    $proplist = "name","cn","lastlogon"
    foreach ($i in $proplist){$objsearcher.PropertiesToLoad.add($i)}
    $results = $objsearcher.findall()
    $serverlist = $

Note, I'm assuming in this example that all DNS servers are also AD DC's and therefore appear in the Domain Controllers OU, if that's not the case for you then you'd need to adjust it accordingly.

$serverlist now contains a list of DNS servers to work with.

From there I can simply use foreach to work through the list, and call the AddDnsServerSecondaryZone cmdlet to add the required secondary as follows :

    foreach ($dnsserver in $serverlist)
        Add-DnsServerSecondaryZone -Name $domain -ZoneFile $zonefile -MasterServers $ipaddr -ComputerName $dnsserver

Since I want this to be runnable at a PowerShell prompt rather than have to edit the script each time, I add the following at the beginning :

    $zonefile=$domain + ".dns"

which reads in the domain and IP Address arguments passed to the script, and then generates the .dns filename to be used in the command (since secondary zones aren't stored in AD).

You can download the completed script, which includes logging and error trapping, from since it also maintains the formatting.

Thursday, 20 June 2013

Adding modules into PowerShell 3.0 ISE for services running on other machines

In my last blog I talked a little about the new Commands menu within the PowerShell 3.0 ISE, the ability to view all the cmdlets relating to an individual module, and then use it to build a command.

It's a great feature and of course there are a lot of modules out there for different applications and services. As you install new roles and features that use their own set of modules the installer adds those modules into the list, so by default the PowerShell ISE modules list contains all the modules relevant to the local installation.

But what if you want to write your script and insert cmdlets into it on a machine that doesn't have those roles or features installed? You could install the relevant management pack, but that seems somewhat overkill to me.

The other option is to simply copy the relevant files from a machine that already has them.

All of the module files installed on a machine can be found in :


with an individual folder for each module installed. These obviously get loaded when PowerShell ISE loads.

To make these available on another machine, for instance making the DNS Server module show up on your desktop machine, simply copy the relevant folder(s) (in this case DnsServer) from a server that it's installed on, into the modules folder on your target machine (your desktop).

The next time you load the PowerShell ISE you'll find the module and its list of cmdlets listed and available for use. This way on one machine you can have all your required modules, be they SQL Server, Exchange Server, Active Directory, DNS Server etc on a single machine.


Just to avoid confusion, my intention when finding this method was simply to allow me to write the code on my local machine, NOT to allow me to execute the cmdlets locally. Using the -Computer parameter to specify the machine to run the command against won't work, since your local machine is missing the required dll's. That said, if you've established a connection to the remote server, eg using Enter-PSSession, then the cmdlets generated on your local machine can be run since you're using the dll's on the remote machine not the local ones.


Wednesday, 19 June 2013

PowerShell v3 and new features including DNS Server cmdlets

With the introduction of PowerShell v3 in Windows Server 2012 we now have a new collection of cmdlets to play with, and amoung them are a collection of cmdlets for controlling and administering DNS Server.

As someone who's been rewriting many old batchscripts to use PowerShell, and is also in the process of migrating to a new 2012 DNS setup this obviously came as great news. No more using PowerShell as a wrapper for dnscmd or having to dig into WMI calls. Unfortunately as is often the case being up to date has its drawbacks, yep, you guessed it, there's a definite drought of documentation out there explaining how to use it all. There are other additions as well, but DNS Server's the area I've been playing with recently.

For a straight forward list of DNS cmdlets check out Once you know which cmdlet you need the easiest option is perhaps to use the Commands menu within the PowerShell v3 ISE, which if not already displayed on the right hand side can be viewed by selecting View and then Show Command Add-on.

Using the new commands menu definitely helps, for any cmdlets you haven't used before. Selecting the DnsServer option from the modules list gives you a list of the available cmdlets in that category, and selecting one of them displays the parameter details for it. Note, where the options change depending on what you're doing (for instance using Add-DnsServerResouceRecord, where A records have different options to MX records etc), you'll see tabs along the top to allow you to select the required set of parameters. Simply fill in the required text boxes, and then click Insert, the complete PowerShell command line using those parameters will be created in the bottom window. Either run it there or copy and paste the code into your script.

If you're unsure which details need to go into which box, the -WhatIf parameter will let you know what your current selection would do if it was run (without actually doing it and potentially doing something wrong / unexpected). The WhatIf parameter isn't new to v3, but its combination and availability in the commands menu makes it even more useful. One thing worth noting about the -WhatIf parameter is that it gives an overview of what will happen, not always the exact detail. Take the following example :

Add-DnsServerResourceRecord -DomainName -Name _autodiscover._tcp -Port 443 -Priority 0 -Srv -Weight 0 -ZoneName -WhatIf

The output produced will be :

What if: Adding DNS resource record _autodiscover._tcp of type SRV in zone on MYSERVER server.

You'll see that it confirms the command will create the resource record, that it's an SRV record, where it is and in which zone, but not the finer details. So if you put the port details in the wrong place then WhatIf won't help.

That's enough for now, the next couple of blogs will be looking at specific DNS Server cmdlets and how they can be used.