some of you maybe encountered following issue during the deployment of the Azure Stack RTM PoC.
Lets look on the field configuration:
One server HP DL360 G8
NIC Type 1GBE Intel i360 (HP OEM Label)
Two Public IPv4 Adresses published directly to the host and host configured as exposed host in the border gateway firewalls
No Firewall Rules for that host on the gateways
Switchports for that host configured as Trunk/Uplink ports with VLAN tagging enabled
We use Azure AD for Authentication
In my case, the important point is the port trunk and the VLAN tagging.
Normally VLAN tagging is no issue because the deployment toolkit should set the tag automatically during deployment for all VMs required and the host system.
In my case and during many test and validation deployments, that didn’t happen. After I start the deployment, a new virtual switch will be deployed and a virtual NIC named “deployment” will be configured for the host. Afterwards the deployment starts. Around 3 hours later, the deployment stops in step 60.120.121 and could not connect to the identity provider.
Whats the reason for the failure?
First you should know, that the Azure Stack Deployment switches between host and BGPNAT VM for internet communication. Mostly all traffic runs through the NAT VM but in that case, the host communicates directly with the internet.
So what happend? After creating the “deployment” NIC for the host, the deployment tool didn’t set the VLAN Tag on that virtual NIC. That breaks the network communication for the host, for the VMs there isn’t any issue because the VLAN is set for the NAT VM correctly.
What is the Workaround?
Start the deployment and configure it like normal
Let the deployment run into the failure
Open a new PowerShell with admin permissions (Run as Administrator)
Data disk drive configuration: All data drives must be of the same type (SAS or SATA) and capacity. If SAS disk drives are used, the disk drives must be attached via a single path (no MPIO, multi-path support is provided)
HBA configuration options: 1. (Preferred)Simple HBA
2. RAID HBA – Adapter must be configured in “pass through” mode
3. RAID HBA – Disks should be configured as Single-Disk, RAID-0
Supported bus and media type combinations
RAID SSD (If the media type is unspecified/unknown*)
SATA SSD + SATA HDD**
SAS SSD + SAS HDD**
* RAID controllers without pass-through capability can’t recognize the media type. Such controllers will mark both HDD and SSD as Unspecified. In that case, the SSD will be used as persistent storage instead of caching devices. Therefore, you can deploy the Microsoft Azure Stack POC on those SSDs.
** For tiered storage, you must have at least 3 HDDs.
Example HBAs: LSI 9207-8i, LSI-9300-8i, or LSI-9265-8i in pass-through mode
While the above configuration is generic enough that many servers should fit the description, we recommend a couple of SKUs: Dell R630 and the HPE DL 360 Gen 9. Both these SKUs have been in-market for some time.
Microsoft offers a new free eBook written by Shreesh Dubey, Vijay Tandra Sistla, Shivam Garg, and Aashish Ramdas. Within the book you can read some really usefull things, how to Backup your Microsoft Hybrid cloud.
Issue 1 The PowerShell cmdlets Get-MgmtSvcSubscription, Get-MgmtSvcPlan and Get-MgmtSvcAddOn cannot retrieve a Subscription, a Plan, or AddOn through a specific Subscription, Plan, or AddOn ID, respectively.
Symptoms When you use these PowerShell cmdlets to retrieve any of these three objects, you notice that the command allows only the retrieval operation through the specific name of the objects, that you cannot retrieve base on a specific ID. Resolution Now the parameter with a corresponding ID can be used to retrieve the corresponding object.
Issue 2 Usage Collector Service loses its configuration after Windows Azure Pack Update 1 is applied.
Symptoms When you deploy Update 1 over an existing deployment by having the Usage Collector Service configured and fully functional, you notice that the Usage Service stops working. Resolution This issue does not occur in Update 2. The Usage Service configuration is preserved and no manual steps are required.
Issue 3 The Plan Creation Event provides NULL for the Plan Display Name to Services registered to receive this notification.
Symptoms For any Service that is registered to receive notifications, the Plan Display Name is passed as NULL when a Plan Creation event is generated. Resolution In Update 2, the Display name value is passed instead of NULL.
Issue 4 The Web Application Gallery Service generates a large amount of temp files in its Temp folder.
Symptoms Every time that the WebAppGallery service restarts, it creates a new temp folder. The temp folder stores the cached feeds that contains the Metadata of the available gallery applications and their dependencies and the actual application packages. The service is scheduled to restart every 1,740 minutes. This causes the growing number of temp files over time. Resolution In Update 2, the Display name value is passed instead of NULL.
Issue 5 Resizing dynamic memory virtual machines through machine profiles causes an error when the new maximum memory exceeds the original maximum memory. Also, changing between static to dynamic or vice versa in the machine profile results in an error.
Symptoms When a tenant user sets the value of the dynamic memory virtual machine to a value that is greater than the original maximum value, an error is generated and the operation fails. Resolution This process is now possible in Update 2 and no errors are generated.
Issue 6 Active Directory was not enabled for Windows Azure Pack WebSites Resource Provider.
Symptoms Administrators had no way to configure WebSites Clouds to use Active Directory. Tenants have no way of enabling Active Directory Account Management and Authentication for their websites. Resolution In Update 2, this functionality is available for Administrators to configure and for Tenants to use.
Issue 7 It is difficult to discover how to create Co-Administrators for an existing subscription.
Symptoms Tenant users typically go to “My Account” and “Subscription” drawer and pane, but do not see an option there to create Co-Administrator. Resolution In this Update, a new link is added to the “My Account” drawer set of options.
Issue 8 Error messages in Administration Site that refer to un-configured services in a Plan do not specify the problems.
Symptoms When you try to update a Plan, such as when you make it public, an error is generated if any of the Resource Providers it contains are not configured. This message does not specify the problem or which Resource Providers are affected. Resolution In Update 2, the messages are more informative, and the user can take corresponding actions.
Issue 9 Subscription Update Events are provided in the Pending Approval state instead of the Acknowledge state.
Symptoms Billing System receives the Subscription Update Event in the Pending Approval state. Resolution In Update 2, the Billing System receives the Subscription Update Event in the correct state.
Issue 10 An obscured error is returned when a Tenant creates a database by using a security token whose corresponding password does not meet the SQL Server domain policies.
Symptoms The user receives an unrelated SQL error about a SELECT statement that contains an invalid column. Resolution In Update 2, the tenant sees the correct error and can resolve it by changing the credential password.