Creating a Cloud Architecture Roadmap

Image result for cloud architecture jpg

Overview

When a product has been proved to be a success and has just come out of a MVP (Minimal Viable Product) or MMP (Minimal Marketable Product) state, usually a lot of corners would have been cut in order to get a product out and act on the valuable feedback. So inevitably there will be technical debt to take care of.

What is important is having a technical vision that will reduce costs and provide value/impact/scaleable/resilient/reliable which can then be communicated to all stakeholders.

A lot of cost savings can be made when scaling out by putting together a Cloud Architecture Roadmap. The roadmap can then be communicate with your stakeholders, development teams and most importantly finance. It will provide a high level “map” of where you are now and where you want to be at some point in the future.

A roadmap is every changing, just like when my wife and I go travelling around the world. We will have a roadmap of where want to go for a year but are open to making changes half way through the trip e.g. An earthquake hits a country we planned to visit etc. The same is true in IT, sometimes budgets are cut or a budget surplus needs to be consumed, such events can affect your roadmap.

It is something that you want to review on a regular schedule. Most importantly you want to communicate the roadmap and get feedback from others.

Feedback from other engineers and stakeholders is crucial – they may spot something that you did not or provide some better alternative solutions.

Decomposition

The first stage is to decompose your ideas. Below is a list that helps get me started in the right direction. This is by no means an exhausted list, it will differ based on your industry.

Component Description Example
Application Run-timeWhere apps are hostedAzure Kubernetes
Persistent StorageNon-Volatile DataFile Store
Block Store
Object Store
CDN
Message
Database
Cache
Backup/RecoveryBackup/Redundant SolutionsManaged Services
Azure OMS
Recovery Vaults
Volume Images
GEO Redundancy
Data/IOTConnected Devices / SensorsStreaming Analytics
Event Hubs
AI/Machine Learning
GatewayHow services are accessedAzure Front Door, NGIX, Application Gateway, WAF, Kubernetes Ingress Controllers
Hybrid ConnectivityOn-Premise Access
Cross Cloud
Express Route
Jumpboxes
VPN
Citrix
Source ControlWhere code lives
Build – CI/CD
Github, Bitbucket
Azure Devops, Octopus Deploy, Jenkins
Certificate ManagementSSL CertificatesAzure Key Vault
SSL Offloading strategies
Secret ManagementStore sensitive configurationPuppet (Hiera), Azure Keyvault, Lastpass, 1Password
Mobile Device ManagementGoogle Play
AppStore
G-Suite Enterprise MDM etc

Once you have an idea of all your components. The next step is to breakdown your road-map into milestones that will ultimately assist in reaching your final/target state. Which of course will not be final in a few years time 😉 or even months!

Sample Roadmap

Below is a link to a google slide presentation that you can use for your roadmap.

https://docs.google.com/presentation/d/1Hvw46vcWJyEW5b7o4Xet7jrrZ17Q0PVzQxJBzzmcn2U/edit?usp=sharing

Advertisements

Query Azure AppInsights with Powershell

In order to query AppInsights using powershell, you will need your AppInsights AppId and APIKey.

The important consideration is to ensure your JSON is valid, so always run it through a parser and use the correct escape characters for both JSON and PowerShell. Have a look at the string in $queryData.

The following code will query appinsights and generate csv files based on the batch size. It also using paging by leveraging:

| serialize | extend rn = row_number()

Happy DevOps Reporting 🙂

param (
[Parameter(Mandatory = $true)]
[string] $AppInsightsId,

[Parameter(Mandatory = $true)]
[string] $apiKey,

[Parameter(Mandatory = $false)]
[string]
$Timespan = "P7D",

[Parameter(Mandatory = $false)]
[int]
$batchSize = "10000",

[Parameter(Mandatory = $false)]
[string]
$OutputFolder = "C:\Output\",

[Parameter(Mandatory = $false)]
[string]$logFileName = "AppQuery.log",

[Parameter(Mandatory = $false)]
[string]$logFolder = "C:\Logs\"
)

Add-Type -AssemblyName System.Web

Import-Module .\Helpers.psm1 -Force -ErrorAction Stop
Import-Module .\Shared.Logging.psm1 -Force -Global
Import-Module .\Security.Helpers.psm1 -Force -Global

function prepareFileHeader($filenumber, $columnNames) {
$csvString = ""
ForEach ( $Property in $columnNames )
{
$csvString += "$($Property.Name),"
}
$csvString = $csvString.Substring(0,$csvString.Length -1)
$file = Join-Path $OutputFolder "batch-$i.csv"
$csvString | Out-File -filepath $file -Encoding utf8
$csvString = $null
}

function writeRecordsToFile($records) {
ForEach ( $record in $records )
{
$csvString = ""
foreach($cell in $record) {
$csvString += "$cell,"
}
$csvString = $csvString.Substring(0,$csvString.Length -1)
$file = Join-Path $OutputFolder "batch-$i.csv"
$csvString | Out-File -filepath $file -Encoding utf8 -Append -NoClobber
$csvString = $null
}
}

$logFilePath = PrepareToLog $logFolder $logFileName

try {
$url = "https://api.applicationinsights.io/v1/apps/$AppInsightsId/query"
$headers = @{"Content-Type" = "application/json"}
$headers.add("x-api-key", $apiKey)
$queryString = "?timespan=$Timespan"
$fullUrl = $url + $queryString

$queryTotalMessageCount = "traces\r\n | where message contains \`"Max Retry Count reached\`" and message contains \`"MessageService\`"\r\n | summarize count()"
$queryTotalMessageCountBody = "{
`"query`": `"$queryTotalMessageCount`"
}"
$resultCount = Invoke-WebRequest -Uri $fullUrl -Headers $headers -Method POST -Body $queryTotalMessageCountBody -ErrorAction Continue
$totalObject = ConvertFrom-Json $resultCount.Content

$totalRecords = $totalObject.tables.rows[0]
$pages = [math]::ceiling($totalRecords/$batchSize)
$startRow = 0
$endRow = $batchSize

Write-Host "Total Files: $pages for Batch Size: $batchSize"
For ($i=1; $i -le $pages; $i++) {
Write-Host "Processing File: C:\batch-$i.csv"
$queryData = "traces\r\n | extend TenantId = extract(\`"Tenant Id \\\\[[a-z0-9A-Z-.]*\\\\]\`", 0, message) | extend UniqueTransactionId = extract(\`"\\\\[[a-z0-9A-Z-. _\\\\^]*\\\\]\`",0 ,extract(\`"Message Transaction \\\\[[a-z0-9A-Z-._]*\\\\]\`", 0, message))\r\n | extend TransactionId = trim_start(\`"\\\\[\`", tostring(split(UniqueTransactionId, \`"_\`") [0]))\r\n | extend TransactionDateTicks = tostring(split(UniqueTransactionId, \`"_\`") [1])\r\n | extend PrincipalId = trim_end(\`"\\\\]\`", tostring(split(UniqueTransactionId, \`"_\`") [2]))\r\n | where message contains \`"Max Retry Count reached\`" and message contains \`"MessageService\`"\r\n | project TransactionId, TransactionDateTicks, PrincipalId, TenantId\r\n | summarize ErrorCount = count(TransactionId) by TransactionId, TransactionDateTicks, PrincipalId, TenantId\r\n | serialize | extend rn = row_number()\r\n | where rn > $startRow and rn <= $endRow"
$queryBody = "{
`"query`": `"$queryData`"
}"
$result = Invoke-WebRequest -Uri $fullUrl -Headers $headers -Method POST -Body $queryBody -ErrorAction Continue
$data = ConvertFrom-Json $result.Content
$startRow += $batchSize
$endRow += $batchSize

if($i -eq 1) {
$columnNames = $data.tables.columns | select name
}

prepareFileHeader $i $columnNames
writeRecordsToFile $data.tables.rows
}
} catch {
LogErrorMessage -msg $error[0] -filePath $logFilePath -fatal $true
}

ARM – Modular Templates – Reference resources already created

Hi,

I noticed the Microsoft documentation related to the following function is a little bit vague.

reference(resourceName or resourceIdentifier, [apiVersion], [‘Full’])

The second issue is see a lot of people having is how do you reference a resource already created in ARM and get some of that objects properties e.g. FQDN on a public IP already created etc.

The clue to solve this issue, so that ARM Template B can reference a resource created in ARM Template A can be found here:

By using the reference function, you implicitly declare that one resource depends on another resource if the referenced resource is provisioned within same template and you refer to the resource by its name (not resource ID). You don’t need to also use the dependsOn property. The function isn’t evaluated until the referenced resource has completed deployment.

Or use linked templates (Linked templates is a huge rework and you need to host the files on the net). Lets see if we can do it via resourceId.

Therefore if we do reference a resource by resourceId, we will remove the implicit “depends on”, allowing ARM Template B to use a resource created in a totally different ARM template.

A great example might be the FQDN on an IP Address.

Imagine ARM Template A creates the IP Address


"resources": [
{
"apiVersion": "[variables('publicIPApiVersion')]",
"type": "Microsoft.Network/publicIPAddresses",
"name": "[variables('AppPublicIPName')]",
"location": "[variables('computeLocation')]",
"properties": {
"dnsSettings": {
"domainNameLabel": "[variables('AppDnsName')]"
},
"publicIPAllocationMethod": "Dynamic"
},
"tags": {
"resourceType": "Service Fabric",
"scaleSetName": "[parameters('scaleSetName')]"
}
}]

Now Imagine we need to get the FQDN of the IP Address in a ARM Template B

What we going to do is try this:

reference(resourceIdentifier, [apiVersion]) ->
reference(resourceId(), [apiVersion]) ->
e.g.

Here is an example where ARM template B references a resource in A and gets a property.


"managementEndpoint": "[concat('https://',reference(resourceId('Microsoft.Network/publicIPAddresses/',variables('AppPublicIPName')), variables('publicIPApiVersion')).dnsSettings.fqdn,':',variables('nodeTypePrimaryServiceFabricHttpPort'))]",

The important thing here is to ensure you always include the API Version. This pattern is a very powerful way to create smaller and more modular ARM templates.

Note: In the above pattern, you do not need to define DependsOn in ARM Template B, as we are explicitly defining a reference to an existing resource. ARM Template B is not responsible for creating a public IP. If you need it, you run ARM Template A.

So if you need a reference to existing resources use the above. If you need a reference to resources created in the SAME ARM template use:

reference(resourceName)

Cheers

Puppet Enterprise – Structure your Hiera Data

Synopsis

This post will discuss how to structure your Hiera Data, so that your profiles will automatically inject the parameters.

Why? So we can keep our profile classes and other classes super clean and succinct.

If you have If else statements in your classed depending on what environment or node the code is running on, you might have a code structure smell. Lets dig in.

Assumptions

You are using a Puppet Control Repository and leveraging Code Manager (R10K) to manage your code with Puppet Enterprise

Secondly you are using Profiles and Roles pattern to structure your classes.

I highly recommend you download the Puppet Control Repository template here.

Profiles and Roles

The most important aspect to consider is structuring your Profiles and Roles to accept parameters that can be resolved and matched to Hiera Data.

Here we have a role for all our Jumpboxes that we can use to remote into.
As we can see it will have the following profiles applied:


class role::jumpbox {
include profile::base
include profile::jumpbox::jumpboxsoftware
include profile::jumpbox::firewall
include profile::jumpbox::hosts
}

Lets pick one of these profiles that require data from Hiera.


class profile::jumpbox::hosts (
  String $hostname = 'changeme',
  String $ip = 'changeme',
)
{
  host { $hostname:
    ensure => present,
    ip     => $ip,
  }
}

The above profile ensures that the /etc/hosts file has some entries in it.

It accepts two parameters:
profile::jumpbox::hosts::ip

profile::jumpbox::hosts::hostname

Similar to Java or C# we can use a sort of dependency injection technique, where puppet will automatically look for this parameter in Hiera; a key/value store.

Hiera

The trick is to structure your Hiera Data and use the same Fully Qualified Names in the keys.

Each environment needs a different set of host names.

I then have the following structure in the control repro

.data\<environment1>\jumpbox\conf.yaml
.data\<environment2>\jumpbox\conf.yaml
.data\<environment3>\jumpbox\conf.yaml

Each folder in data represents an Environment in Puppet Classifications:

The second important convention is we use a geography variable in each Environment to resolve Hiera Data automatically.

Go to your Puppet Master Enterprise Web Console and manage the Classifications.

What you are doing is creating a variable that can be used by the hiera.yaml file to dynamically load data for the correct environment when the agent runs.

On the Puppet Master we need to setup our environments to match the Control Repository and add the magic variable. Any Node that runs the puppet agent will then have this variable set. This can then be used to load the corresponding Hiera config file.

Here we can see Environment1 has a variable defined called geography that matches the Environment name. We can then leverage this convention:

Puppet Profile -> Hiere Data lookup -> Folder that matches the variable name -> resolve parameter

This is all done automatically for you.

Puppet Control Repository Structure

The repository then looks like this:

Let us dig a little deeper and see how this structure is configured.

hiera.yaml

.\hiera.yaml

This file now contains the instructions to tell Hiera how to load our data.

—hiera.yaml—


---
version: 1

defaults:
  datadir: "data"

hierarchy:
  - name: 'Yaml Key Value Store'
    data_hash: yaml_data
    paths:
      - "%{geography}/jumpbox/conf.yaml"
      - "common.yaml"

  - name: "Encrypted Data"
    lookup_key: eyaml_lookup_key
    paths: 
      - "%{geography}/jumpbox/secrets.eyaml"
      - "common.eyaml"
    options:
      pkcs7_private_key: /etc/puppetlabs/puppet/eyaml/private_key.pkcs7.pem
      pkcs7_public_key: /etc/puppetlabs/puppet/eyaml/public_key.pkcs7.pem

Data – yaml

the .yaml files will contain the same variable names (fully qualified) that match the PROFILE files e.g.

—conf.yaml—


profile::jumpbox::hosts::hostname: 'rdp.rangerrom.com'
profile::jumpbox::hosts::ip: 8.8.8.8'

As you can see above. As long as your profiles and parameters match, Hiera will automatically inject the correct parameter for each environment.

Hiera will resolve – %{geography}/jumpbox/conf.yaml

In the Puppet master you had setup your classifications, so when the puppet agent runs on Environment1 nodes, it will get the jumpbox/conf.yaml that matches the variable name geography=”Environment1″

Encrypted Data – eyaml

Encrypted data is just as easy to store.
* Generate the encrypted data.
* Store the data in an eyaml file in the same folder as the yaml data.
* Add a path to the data in the hiera.yaml file.

We have encrypted data e.g. the default local admin account setup via the profile – include profile::base
We use the Puppet Master private key to generate the encrypted data, see the end of this blog on how to create encrypted data.

—secrets.eyaml—


profile::base::adminpassword: >
    ENC[PKCS7,MIIBeQYJKoZIhvcNAQcDoIIBajCCAWYCAQAxggEhMIIBHQIBADAFMAACAQEw
    DQYJKoZIhvcNAQEBBQAEggEAnMWlddVoU9lC8tBNvOLI9OYI6xtCD0y3NIVe
    Ylm25dUZ8sqGP+yVQ8Y0P5xIse5f/WVOkavByZJK5yV4fDYFpD6IhXk4IJUe
    dVUw8VmO/RG84AknDDrtNPlSPm4uQqYPOOa0BmgO1iiOY4rcAxhFzT5nzod3
    MIK7lmbuP859R5jtJ5PZxZKCNERGY+dxUZfcdPs0/zr/KgLGcHc/awzYtEuI
    0tOGPp80gTVkhmCHO7KuClsg97XTRGi0BfiuiyjOWLIeAx5hbhMHi65ZPl5U
    MlJFoTA1nw3ATcC6NL3ikECWaQrt2xyxZ1uoYKqvN0ClsFLIqBQ1gXRTvQPD
    SlBQqDA8BgkqhkiG9w0BBwEwHQYJYIZIAWUDBAEqBBCWLuT77kT6q/ojfjKx
    wk17gBATvEM58mGyP5CGbMqlbEip]

How to Encrypt Data

SSH into the Puppet Master. Locate your Puppet Master Certificates. Then run the following


puppetmaster@rangerrom.com:~$ sudo /opt/puppetlabs/puppet/bin/eyaml encrypt -p --pkcs7-public-key=/etc/puppetlabs/puppet/eyaml/public_key.pkcs7.pem

Enter password: ***
string: ENC[PKCS7,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]

OR

block: >
    ENC[PKCS7,MIIBeQYJKoZIhvcNAQcDoIIBajCCAWYCAQAxggEhMIIBHQIBADAFMAACAQEw
    DQYJKoZIhvcNAQEBBQAEggEAnMWlddVoU9lC8tBNvOLI9OYI6xtCD0y3NIVe
    Ylm25dUZ8sqGP+yVQ8Y0P5xIse5f/WVOkavByZJK5yV4fDYFpD6IhXk4IJUe
    dVUw8VmO/RG84AknDDrtNPlSPm4uQqYPOOa0BmgO1iiOY4rcAxhFzT5nzod3
    MIK7lmbuP859R5jtJ5PZxZKCNERGY+dxUZfcdPs0/zr/KgLGcHc/awzYtEuI
    0tOGPp80gTVkhmCHO7KuClsg97XTRGi0BfiuiyjOWLIeAx5hbhMHi65ZPl5U
    MlJFoTA1nw3ATcC6NL3ikECWaQrt2xyxZ1uoYKqvN0ClsFLIqBQ1gXRTvQPD
    SlBQqDA8BgkqhkiG9w0BBwEwHQYJYIZIAWUDBAEqBBCWLuT77kT6q/ojfjKx
    wk17gBATvEM58mGyP5CGbMqlbEip]
puppetmaster@rangerrom.com:~$

 

Puppet – Join Ubuntu 16.04 Servers to an Azure Windows Active Directory Domain

We use Azure Active Directory Domain Services and wanted a single sign on solution for Windows and Linux. The decision was made to join all servers to the Windows Domain in addition to having SSH Key auth.

I assume you know how to use Puppet Code Manager and have a Puppet Control Repository to manage Roles and Profiles.

Ensure you have a Active Directory Service Account that has permissions to join a computer to a domain. You can store this user in Hiera.

The module we need is called Realmd, however the current version (Version 2.3.0 released Sep 3rd 2018) does not support Ubuntu  16.04. So I have a forked repro here that you can use.

Modules

puppetfile


mod 'romiko-realmd',
  :git    => 'https://github.com/Romiko/realmd.git',
  :branch => 'master'

mod 'saz-resolv_conf',                  '4.0.0'

linuxdomain.pp


class profile::domain::linuxdomain (
  String $domain = 'RANGERROM.COM',
  String $user = 'romiko.derbynew@rangerrom.com',
  String $password = 'info',
  Array  $aadds_dns = ['10.0.103.36','10.0.103.37']
) {

  $hostname = $trusted['hostname']
  $domaingroup = downcase($domain)

  host { 'aaddshost':
    ensure  => present,
    name    => "${hostname}.${domain}",
    comment => 'Azure Active Directory Domain Services',
    ip      => '127.0.0.1',
  }

  class { 'resolv_conf':
    nameservers => $aadds_dns,
    searchpath  => [$domain],
  }

  exec { "waagent":
    path        => ['/usr/bin', '/usr/sbin', '/bin'],
    command     => "waagent -start",
    refreshonly => true,
    subscribe   => [File_line['waagent.conf']],
  }

  exec { "changehostname":
    path        => ['/usr/bin', '/usr/sbin', '/bin'],
    command     => "hostnamectl set-hostname ${hostname}.${domain}",
    refreshonly => true,
    subscribe   => [File_line['waagent.conf']],
  }

  file_line { 'waagent.conf':
    ensure             => present,
    path               => '/etc/waagent.conf',
    line               => 'Provisioning.MonitorHostName=y',
    match              => '^Provisioning.MonitorHostName=n',
    append_on_no_match => false,
  }

  class { 'ntp':
    servers => [ $domain ],
  }

  class { '::realmd':
    domain                => $domain,
    domain_join_user      => $user,
    domain_join_password  => $password
  } ->

  file_line { 'sssd.conf':
  ensure             => present,
  path               => '/etc/sssd/sssd.conf',
  line               => '#use_fully_qualified_names = True',
  match              => '^use_fully_qualified_names',
  append_on_no_match => false,
  }

  file_line { 'common-session':
  path => '/etc/pam.d/common-session',
  line => 'session required        pam_mkhomedir.so skel=/etc/skel/ umask=0077',
  }

  file_line { 'sudo_rule':
  path => '/etc/sudoers',
  line => "%aad\ dc\ administrators@${domaingroup} ALL=(ALL) NOPASSWD:ALL",
  }
}

The above code ensures:

  • Its fqdn is in the /etc/hosts file so it can resolve to itself.
  • The Active Directory DNS servers are in resolv.conf and the default search domain is set. This allows nslookup to <computername> to work as well as <computername>.<domainname>
  • Renames the Linux Server and sets Azure to monitor host name changes
  • Sets the Network Time Protocol to use the Active Directory Severs
  • Uses Realmd 
    • Join the domain
    • configure Sssd, samba etc
    • We using my fork until the pull request is accepted
  • Makes changes to SSSD and PAM to ensure smooth operations with Azure Active Directory Domain Services
  • Adds the administrators from AADDS to the sudoers

 

With the above running on your Linux agent, you will have Linux machines using the domain and can leverage single sign on.

CASE SENSITIVE –  Ensure you use Upper Case for $domain.

My source of inspiration to do this configuration came from a Microsoft Document here:

https://docs.microsoft.com/en-us/azure/active-directory-domain-services/active-directory-ds-join-ubuntu-linux-vm

However, here is a Hiera config file to get you going for this class if you like using a key value data source.

common.yaml

#AADDS Domain must be Upper Case else Kerberos Tickets fail
profile::domain::linuxdomain::domain: 'RANGERROM.COM'
profile::domain::linuxdomain::user: 'serviceaccounttojoindomain@RANGERROM.COM'
profile::domain::linuxdomain::aadds_dns:
        - '10.0.103.36'
        - '10.0.103.37'

common.eyaml

profile::domain::linuxdomain::password:
    ENC[PKCS7,MIIBiQ.....==]

In my next article, I will write about setting up a flexible Puppet Control Repository and leveraging Hiera.

Now I can login to the server using my Azure Active Directory credentials that I even use for Outlook, Skype and other Microsoft Products.

ssh -l romiko.derbynew@rangerrom.com myserver   (resolv.conf has a search entry for rangerrom.com, so the suffix will get applied)
ssh -l romiko.derbynew@rangerrom.com myserver.rangerrom.com

Cheers

Azure Database for PostgreSQL – Backup/Restore

PSQL has some awesome tools pg_dump and pg_restore that can assist and cut down restore times in the event of a recovery event.

Scenario

User error – If a table was accidentally emptied at noon today:

• Restore point in time just before noon and retrieve the missing table and data from that new copy of the server (Azure Portal or CLI) , Powershell not supported yet.
• Update firewall rules to allow traffic internally on the new SQL Instance

Then log onto a JumpBox in the Azure cloud.
• Dump the restored database to file (pg_dump)


pg_dump -Fc -v -t mytable -h Source.postgres.database.azure.com -U pgadmin@source -d mydatabase > c:\Temp\mydatabase .dump

The switch -Fc – gives us flexibility with pg_restore later (Table Level Restore)

• Copy the table back to the original server using  pg_restore

You may need to truncate the target table before, as we doing data only restore (-a)
Table Only (not schema):


pg_restore -v -A -t mytable -h Target.postgres.database.azure.com -p 5432 -U pgadmin@target -d mydatabase "c:\Temp\mydatabase .dump"

note: Use a VM in the Azure cloud that has access to SQL via VNET Rules or SQL Firewalls.

We currently use PSQL 10. There is a known issue with VNet Rules. Just contact Microsoft if you need them, else use firewall rules for now.

Issues

VNET Rules and PostgreSQL 10

Issue Definition: If using VET rules to control access to the PostgreSQL 10.0 server, customer gets the following error: psql ‘sslmode=require host=server .postgres.database.azure.com port=5432 dbname=postgres’ –username=pgadmin@server psql: FATAL: unrecognized configuration parameter ‘connection_Vnet’ This is a known issue: https://social.msdn.microsoft.com/Forums/azure/en-US/0e99fb68-47fd-4053-a8be-5f8b87b3a660/azure-database-for-postgresql-vnet-service-endpoints-not-working?forum=AzureDatabaseforPostgreSQL

DNS CNAME

nslookup restoretest2.postgres.database.azure.com

Non-authoritative answer:

Name:    cr1.westus1-a.control.database.windows.net

Address:  23.9.34.71

Aliases:  restoretest2.postgres.database.azure.com

nslookup restoretest1.postgres.database.azure.com

Non-authoritative answer:

Name:    cr1.westus1-a.control.database.windows.net

Address:  23.9.34.71

Aliases:  restoretest1.postgres.database.azure.com

The Host is the same server. The way the connection string works is on the USERNAME.

These two command will connect to the SAME Server instance


pg_restore -v -a -t mytable -h restoretest2.postgres.database.azure.com -p 5432 -U pgadmin@restoretest1 -d mydatabase  "c:\Temp\mydatabase.dump"


pg_restore -v -a -t mytable -h restoretest1.postgres.database.azure.com -p 5432 -U pgadmin@restoretest1 -d mydatabase   "c:\Temp\mydatabase.dump"

The hostname just gets us to the Microsoft PSQL server, it is the username that points us to the correct instance. This is extremely important when updating connection strings on clients!

Happy Life – Happy Wife

Windows Azure – Restore Encrypted VM – BEK and KEK

When restoring an Encrypted Virtual Machine that is managed by Windows Azure, you will need to use PowerShell.

This is a two stage process.

Restore-BackupImageToDisk … | Restore-EncryptedBekVM …

Stage 1

Retrieve the Backup Container – This will contain the encrypted disks and json files with the secret keys.

  • keyEncryptionKey
  • diskEncryptionKey

This will allow you to retrieve the disks and JSON metadata in step 2.

Restore-BackupImageToDisk.ps1

param(
    [Parameter(Mandatory=$true)]
    [string]
    $SubscriptionName="RangerRom",

    [Parameter(Mandatory=$true)]
    [string]
    $ResourceGroup="Ranger-Rom-Prod",

    [Parameter(Mandatory=$true)]
    [string]
    $StorageAccount="RangerRomStorage",

    [Parameter(Mandatory=$true)]
    [string]
    $RecoveryVaultName="RecoveryVault",

    [Parameter(Mandatory=$true)]
    [string]
    $VMName,

    [Parameter(Mandatory=$true)]
    [datetime]
    $FromDate
)

Login-AzureRmAccount
Get-AzureRmSubscription -SubscriptionName $SubscriptionName | Select-AzureRmSubscription
$endDate = Get-Date

$namedContainer = Get-AzureRmRecoveryServicesBackupContainer  -ContainerType "AzureVM" -Status "Registered" | Where { $_.FriendlyName -eq $VMName }
$backupitem = Get-AzureRmRecoveryServicesBackupItem -Container $namedContainer  -WorkloadType "AzureVM"

$rp = Get-AzureRmRecoveryServicesBackupRecoveryPoint -Item $backupitem -StartDate $FromDate.ToUniversalTime() -EndDate $enddate.ToUniversalTime()
$restorejob = Restore-AzureRmRecoveryServicesBackupItem -RecoveryPoint $rp[0] -StorageAccountName $StorageAccount -StorageAccountResourceGroupName $ResourceGroup
Wait-AzureRmRecoveryServicesBackupJob -Job $restorejob -Timeout 43200
$restorejob = Get-AzureRmRecoveryServicesBackupJob -Job $restorejob
$details = Get-AzureRmRecoveryServicesBackupJobDetails -Job $restorejob
$details

Stage 2

We will grab the job details via the pipeline if provided or go find the earliest Restore Job after the FromDate and then initiate a restore of all the encrypted disks to the new Virtual Machine.

Restore-EncryptedBekVM.ps1

[CmdletBinding()]
param(
    [Parameter(ValueFromPipeline=$True, Mandatory=$false, HelpMessage="Requires a AzureVmJobDetails object e.g. Get-AzureRmRecoveryServicesBackupJobDetails. Optional.")]
    $JobDetails,

    [Parameter(Mandatory=$true, HelpMessage="Assumes osDisks, DataDisk, RestorePoints, AvailbilitySet, VM, Nic are all in same resource group.")]
    [string]
    $DefaultResourceGroup="Ranger-Rom-Prod",
    
    [Parameter(Mandatory=$true)]
    [string]
    $SourceVMName="Lisha",

    [Parameter(Mandatory=$true)]
    [string]
    $TargetVMName,

    [Parameter(Mandatory=$true)]
    [string]
    $BackupVaultName="RecoveryVault",

    [Parameter(Mandatory=$true)]
    [datetime]
    $FromDate
)

Begin {
    $FromDate = $FromDate.ToUniversalTime()
    Write-Verbose "Started. $(Get-Date)"
}

Process {
	Write-Verbose "Retrieving Backup Vault."
	if(-not $JobDetails)
	{
		Get-AzureRmRecoveryServicesVault -Name $BackupVaultName -ResourceGroupName $DefaultResourceGroup | Set-AzureRmRecoveryServicesVaultContext
		$Job = Get-AzureRmRecoveryServicesBackupJob -Status Completed -Operation Restore -From $FromDate | Sort -Property StartTime | Where { $_.WorkloadName -eq $SourceVMName} | Select -Last 1
		if($Job -eq $null) {
			throw "Job $workLoadName not found."
		}
		$JobDetails = Get-AzureRmRecoveryServicesBackupJobDetails -Job $Job
	}

	Write-Verbose "Query the restored disk properties for the job details."
	$properties = $JobDetails.properties
	$storageAccountName = $properties["Target Storage Account Name"]
	$containerName = $properties["Config Blob Container Name"]
	$blobName = $properties["Config Blob Name"]

	Write-Verbose "Found Restore Blob Set at $($Properties['Config Blob Uri'])"
	Write-Verbose "Set the Azure storage context and restore the JSON configuration file."

	Set-AzureRmCurrentStorageAccount -Name $storageaccountname -ResourceGroupName $DefaultResourceGroup
	$folder = New-Item "C:\RangerRom" -ItemType Directory -Force
	$destination_path = "C:\$($folder.Name)\$SourceVMName.config.json"
	Get-AzureStorageBlobContent -Container $containerName -Blob $blobName -Destination $destination_path
	Write-Verbose "Restore config saved to file. $destination_path"
	$restoreConfig = ((Get-Content -Path $destination_path -Raw -Encoding Unicode)).TrimEnd([char]0x00) | ConvertFrom-Json

	# 3. Use the JSON configuration file to create the VM configuration.
	$oldVM = Get-AzureRmVM | Where { $_.Name -eq $SourceVMName }
	$vm = New-AzureRmVMConfig -VMSize $restoreConfig.'properties.hardwareProfile'.vmSize -VMName $TargetVMName -AvailabilitySetId $oldVM.AvailabilitySetReference.Id
	$vm.Location = $oldVM.Location

	# 4. Attach the OS disk and data disks - Managed, encrypted VMs (BEK only
	$bekUrl = $restoreConfig.'properties.storageProfile'.osDisk.encryptionSettings.diskEncryptionKey.secretUrl
	$keyVaultId = $restoreConfig.'properties.storageProfile'.osDisk.encryptionSettings.diskEncryptionKey.sourceVault.id
	$kekUrl = $restoreConfig.'properties.storageProfile'.osDisk.encryptionSettings.keyEncryptionKey.keyUrl
	$storageType = "StandardLRS"
	$osDiskName = $vm.Name + "_Restored_" + (Get-Date).ToString("yyyy-MM-dd-hh-mm-ss") + "_osdisk"
	$osVhdUri = $restoreConfig.'properties.storageProfile'.osDisk.vhd.uri
	$diskConfig = New-AzureRmDiskConfig -AccountType $storageType -Location $restoreConfig.location -CreateOption Import -SourceUri $osVhdUri
	$osDisk = New-AzureRmDisk -DiskName $osDiskName -Disk $diskConfig -ResourceGroupName $DefaultResourceGroup
	Set-AzureRmVMOSDisk -VM $vm -ManagedDiskId $osDisk.Id -DiskEncryptionKeyUrl $bekUrl -DiskEncryptionKeyVaultId $keyVaultId -KeyEncryptionKeyUrl $kekUrl -KeyEncryptionKeyVaultId $keyVaultId -CreateOption "Attach" -Windows

	$count = 0
	foreach($dd in $restoreConfig.'properties.storageProfile'.dataDisks)
	{
		$dataDiskName = $vm.Name + "_Restored_" + (Get-Date).ToString("yyyy-MM-dd-hh-mm-ss") + "_datadisk" + $count ;
		$dataVhdUri = $dd.vhd.uri;
		$dataDiskConfig = New-AzureRmDiskConfig -AccountType $storageType -Location $restoreConfig.location -CreateOption Import -SourceUri $dataVhdUri
		$dataDisk = New-AzureRmDisk -DiskName $dataDiskName -Disk $dataDiskConfig -ResourceGroupName $DefaultResourceGroup
		Add-AzureRmVMDataDisk -VM $vm -Name $dataDiskName -ManagedDiskId $dataDisk.Id -Lun $dd.Lun -CreateOption "Attach"
		$count += 1
	}

	Write-Verbose  "Setting the Network settings."

	$oldNicId = $oldVM.NetworkProfile.NetworkInterfaces[0].Id
	$oldNic = Get-AzureRmNetworkInterface -Name $oldNicId.Substring($oldNicId.LastIndexOf("/")+ 1) -ResourceGroupName $DefaultResourceGroup
	$subnetItems =  $oldNic.IpConfigurations[0].Subnet.Id.Split("/")
	$networkResourceGroup = $subnetItems[$subnetItems.IndexOf("resourceGroups")+1]
	$nicName= $vm.Name + "_nic_" + (New-Guid).Guid
	$nic = New-AzureRmNetworkInterface -Name $nicName -ResourceGroupName $networkResourceGroup -Location $oldNic.location -SubnetId $oldNic.IpConfigurations[0].Subnet.Id
	$vm=Add-AzureRmVMNetworkInterface -VM $vm -Id $nic.Id
	$vm.DiagnosticsProfile = $oldVM.DiagnosticsProfile

	Write-Verbose "Provisioning VM."
	New-AzureRmVM -ResourceGroupName $oldVM.ResourceGroupName -Location $oldVM.Location -VM $vm

}

End {
    Write-Verbose "Completed. $(Get-Date)"
}


 Usage

.\Restore-BackupImageToDisk … | .\Restore-EncryptedBekVM.ps1 …

Summary

Ensure you have Azure VM Backups and they are backed on on a regular schedule.

Use this script to backup a Restore Point from a Recovery Vault Container.

Remember it will pick the earliest date of a Restore Point relative to your From Date. If there are 4 Restores in the recovery vault from 1 May, it will pick the first one.

Remember to:

  • Decommission the old VM
  • Update DNS with the new ip address
  • Update the Load Balance Sets (If using a Load Balancer)

Update Management solution in Azure – Workspaces

Problem

Update Management in Azure does not support system workspace. If you are using Security Center in Azure, the chances are; all your VM’s are allocated to the system workspace that Security Center created.

The selected workspace is a system workspace and cannot be linked to this account

 

The other error you can get is when you try enable Update Management on VM, and that VM is in another Workspace.

The selected Automation account is already linked to a Long Analytics workspace that is not the selected workspace

Solution

Deallocate All VM’s that you want to use Update Management from the System Workspace to a New Workspace.

  1.  In OMS enable Update Management and Create a New Workspace
  2. Open the Workspace associated with Update Management and note the
    1. Workspace ID
    2. Workspace Primary Key
  3.  Run the following Powershell Scripts
    1. Disconnect-AllVirtualMachinesFromWorkspace
    2. Connect-AllVirtualMachinesToWorkspace -omsID “12345…” -omsKey  “12345…==”
function Disconnect-AllVirtualMachinesFromWorkspace {
        $typeWin = "MicrosoftMonitoringAgent"
        $typeLin = "OmsAgentForLinux"

		$windows = Get-AzureRmVm  | Where { $_.StorageProfile.OsDisk.OsType -eq "Windows" }
		foreach ($vm in $windows) {
			Remove-AzureRmVMExtension -ResourceGroupName "$($vm.ResourceGroupName)" -Name $typeWin -VMName "$($vm.Name)" -Force
        }

        $linux = Get-AzureRmVm  | Where { $_.StorageProfile.OsDisk.OsType -eq "Linux" }-Force
		foreach ($vm in $linux) {
			Remove-AzureRmVMExtension -ResourceGroupName "$($vm.ResourceGroupName)" -Name $typeLin -VMName "$($vm.Name)" -Force
		}
}

function Connect-AllVirtualMachinesToWorkspace {
    param(
        [Parameter(Mandatory=$true, HelpMessage="Workspace Id")]
        [string]
        $omsId,

        [Parameter(Mandatory=$true, HelpMessage="Workspace Primary Key")]
        [string]
        $omsKey
    )

    $typeWin = "MicrosoftMonitoringAgent"
    $typeLin = "OmsAgentForLinux"

    $PublicSettings = New-Object psobject | Add-Member -PassThru NoteProperty workspaceId $omsId | ConvertTo-Json
    $protectedSettings = New-Object psobject | Add-Member -PassThru NoteProperty workspaceKey $omsKey | ConvertTo-Json

    $windows = Get-AzureRmVm  | Where { $_.StorageProfile.OsDisk.OsType -eq "Windows" }
    foreach ($vm in $windows) {
        Set-AzureRmVMExtension -ExtensionName $typeWin -ResourceGroupName  "$($vm.ResourceGroupName)" -VMName "$($vm.Name)" -Publisher "Microsoft.EnterpriseCloud.Monitoring" -ExtensionType $typeWin -TypeHandlerVersion 1.0 -SettingString $PublicSettings  -ProtectedSettingString $protectedSettings  -Location $vm.location
    }

    $linux = Get-AzureRmVm  | Where { $_.StorageProfile.OsDisk.OsType -eq "Linux" }-Force
    foreach ($vm in $linux) {
        Set-AzureRmVMExtension -ExtensionName $typeLin -ResourceGroupName  "$($vm.ResourceGroupName)" -VMName "$($vm.Name)" -Publisher "Microsoft.EnterpriseCloud.Monitoring" -ExtensionType $typeLin -TypeHandlerVersion 1.0 -SettingString $PublicSettings  -ProtectedSettingString $protectedSettings  -Location $vm.location
    }
}

Octopus Deploy – AWS EC2 Web Servers and Elastic Load Balancer

Octopus Deploy has great add on template steps.  Using the

  • AWS – Register ELB Instance
  • AWS – Deregister ELB Instance

Octopus Steps Template

You can then leverage Environments to create Web Server Groups and thus automate deploying your web server instances.

  1. The first step is for Development and Staging, which do not use an ELB
  2. The next 4 steps is for Web Servers in Group 1
  3. The last 4 steps is for Web Servers in Group 2

A Powershell script is used to poll the web server, after the NUGET package is used to deploy the website. It will poll the Web Servers in the group until the homepage returns an HTTP 200 code for all of them. The homepage does various calls to the database, so this page is great to use. You could also create a custom up-time endpoint to query. A really cool way, is to have an API endpoint that returns which servers are up and use this to validate the step for polling.

 

 

Installing Puppet Enterprise on CentOS 7 in AWS EC2 with custom public HostName

Hey,

I ran into a few issues when I wanted to install Puppet Enterprise 2-17 in AWS as an EC2 instance. The main issues were

Summary

  • Need to use hostnamectl and cloud.cfg to change my hostname, as I wanted puppet on a public address, not private address, just for a POC
  • I was using a t2.nano and t2.micro, which will not work with Puppet Enterprise 2017 (puppet-enterprise-2017.2.2-el-7-x86_64). The error you get is just Failed to run PE Installer…… So I used a t2.medium to get around the issue.
  • The usual /etc/hosts file needs some settings and DNS registration (Route53 for me)
  • Disabled SELinux (We usually use a VPN)
  • Configure security groups and have 4433 as backup port (Probably not needed)

Preliminary Install Tasks

  1. Get the latest image from CentOS 7 (x86_64) – with Updates HVM
  2. Spin up an instance with at least 4GB memory, I had a lot of installation issues with applying the catalog with low memory. T2.Medium should work. Bigger is better!
    [puppet.rangerrom.com] Failed to run PE installer on puppet.rangerrom.com.
  3. If you not using a VPN then ensure you setup an elastic IP mapped to the instance for the public DNS name
    ElasticIP.PNG
  4. Register the hostname and elastic IP in DNS
    DNS.PNG
  5. Add you hostnames to /etc/hosts (Important!), note I also added puppet as this is the default for installs. This is a crucial step, so make sure you add your hostnames that you want to use. Put the public hostname first. As this is our primary hostname127.0.0.1  puppet.rangerrom.com puppet localhost
  6. Change the hostname of your EC2 Instance. We need to do the following

    #hostnamectl
    #sudo hostnamectl set-hostname puppet.rangerrom.com –static
    #sudo vi /etc/cloud/cloud.cfg

  7. Add the following to the end of cloud.cfg
    preserve_hostname: true
  8. This is the error I got when I first installed puppet (Due to low memory), therefore we will add port 4433 as well to the AWS security in the next step. I think this was due to insufficient memory, so use a T2.Medium instance size, so you have a minimum of 4GB of memory, else java kills itself. However I add it as a backup here in case you run some other service on 443.

    #sudo vi /var/log/puppetlabs/installer/2017-08-08T02.09.32+0000.install.log

    Failed to apply catalog: Connection refused – connect(2) for “puppet.rangerrom.com” port 4433

  9. Create a security group with the following ports open and also do the same for the Centos Firewall.
    PuppeSecurityGroups
  10. Run  netstat -anp | grep tcp to ensure no port conflicts.
  11. Disable SELinux or have it configured to work in a Puppet Master environment. Edit

    #sudo vi /etc/sysconfig/selinux

    set
    SELINUX=disabled

  12. Edit the sudo vi /etc/ssh/sshd_config and enable Root Logins
    PermitRootLogin yes
  13. Download Puppet Enterprise

    #curl -O https://s3.amazonaws.com/pe-builds/released/2017.2.2/puppet-enterprise-2017.2.2-el-7-x86_64.tar.gz
    #tar -xvf puppet-enterprise-2017.2.2-el-7-x86_64.tar.gz

  14. Install NC and use it to test if your ports are accessible.
    sudo yum install nc
    nc -nlvp 3000 (Run in one terminal) 
  15. nc puppet 3000 ( Run from another terminal)
    NC Test Firewalls.PNG
    This is a great way to ensure firewall rules are not restricting your installation. Secondly we testing that the local server can resolve itself, as it is important that you can resolve puppet and also your custom FQDN before running PE install.
  16. Reboot and run hostnamectl, the new hostname should be preserved.

    #sudo hostnamectl set-hostname puppet.rangerrom.com –static
    [centos@ip-172-31-13-233 ~]$ hostnamectl
    Static hostname: puppet.rangerrom.com
    Transient hostname: ip-172-31-13-233.ap-southeast-2.compute.internal
    Icon name: computer-vm
    Chassis: vm
    Machine ID: 8bd05758fdfc1903174c9fcaf82b71ca
    Boot ID: 0227f164ff23498cbd6a70fb71568745
    Virtualization: xen
    Operating System: CentOS Linux 7 (Core)
    CPE OS Name: cpe:/o:centos:centos:7
    Kernel: Linux 3.10.0-514.26.2.el7.x86_64
    Architecture: x86-64

Installation

  1. Now that we done all our preinstall checks, kick off the installer.

    #sudo ./puppet-enterprise-installer

  2. Enter 1 for a guided install.
  3. Wait until it asks you to connect to the server on https://<fqdn&gt;:3000
    This is what occurs if you did not configure your hostname correctly and you want a public hostname (EC2 internal is default):
    PuppetInstallStage1.PNG

    We want our public hostname.
    PuppetInstallStage1Correct
    Puppet will basically run a thin web server to complete the installation with the following command:
    RACK_ENV=production /opt/puppetlabs/puppet/share/installer/vendor/bundler/bin/thin start –debug -p 3000 -a 0.0.0.0 –ssl –ssl-disable-verify &> /dev/null

  4. Recall, we have the above FQDN in our host file, yours will be your hostname that you setup.
  5. Visit your Puppetmaster site at https://fqdn:3000
  6. Ensure in DNS Alias, you add puppet and all other DNS names you want to use. Otherwise the installation will fail.

    You should see the correct default hostname, if not, you got issues…. I added some alias names such as puppet and my internal and external ec2 addresses.

    PuppetWebDNSAlias.PNG

  7. Set an Admin password and click next
  8. Check and double check the settings to confirm.
    PuppetConfirm.PNG
  9. Check the validation rules, since this is for testing, I am happy with the warnings. It would be awesome if puppetlabs did DNS name resolution validation checks on the HostName. Anyways, here we get a warning about memory, 4GB is what is needed, so if you have install failures it may be due to memory!
    Validator.PNG
  10. I am feeling lucky, lets try with 3533MB of RAM 🙂SuccessInstall.PNG