locked
Warning: "The first Fabric upgrade must specify both the code and config versions" RRS feed

 > 

  • Question

  • Question
    Sign in to vote
    1
    Sign in to vote

    I was able to get the installer to install Service Fabric on my local VMs today.  

    After the install fired up the Service Fabric Explorer to check it out.  I was greeted with warnings an errors!

    First I got a warning.  It is in the fabric:/System/UpgradeOrchestrationService and under that PartitionId=00000000-0000-0000-0000-000000006000 (what ever that is).

    The warning message is this: The first Fabric upgrade must specify both the code and config versions

    Then after a while, this goes away and is replaced by an error:

    Error event: SourceId='System.FM', Property='State'.
    Partition is in quorum loss.

    I have googled this and the only references to the warning message is to pre-GA releases.  I went to the "Details" page in the Explorer and it says:

    Code Version  : 5.6.220.9494 
    Config Version: 0

    So I don't think I am on a pre-GA release, but I don't know what config version 0 is about.

    I am at a loss for how to fix this.  Any advice of a next step would be great!

    Note: Here are the links for the full text of the warning and error.




    • Edited by Vaccanoll Thursday, June 8, 2017 10:24 PM
    Thursday, June 8, 2017 10:22 PM

Answers

  • Question
    Sign in to vote
    2
    Sign in to vote

    Update 2017-07-19:

    While setting up another ASF-Cluster On-Premise I encountered the same issues again. ("The first Fabric upgrade must specify both the code and config versions")

    This time i used the correct (latest) API-Version value (04-2017) in the cluster-config right away.

    Since the Error occured again, I revisited the access rights for the network folder which the diagnostic-store string in the cluster-config points to. I realized, that I forgot to give permissons to the gMSA used by this Windows-Security secured cluster. After fixing these permissions, and the known uninstall-reinstall procedure of the ASF-Cluster the problem seemes fixed.

    Reading the Azure-Service Fabric Events in the Windows Event Log (Application- and Service Protocolls -> Microsoft-Service Fabric -> Admin) was very helpful detecting the problem ;)

    Original Post:

    Hi Vaccanoll,

    I encountered the same problems after setting up a new SF On-Prem Cluster on 3 machines.

    Didn't find any good solutions until I realized, that the Configuration-Examples on Github had been updated for the CodePackageVersion 5.6.xxx. See: https://github.com/Azure-Samples/service-fabric-dotnet-standalone-cluster-configuration/tree/master/Samples/PreviousReleases

    I moved forward making 2 changes to my initial used ClusterConfig for the CreateServiceFabricCluster-Script.

    1. I changed the Api-Version like this: "apiVersion": "04-2017" -> See 5.6 Example-Configs

    2. I changed my diagnosticsStore connectionString-property to an accessible, local path instead of a network-drive like this: "connectionstring": "C:\\ProgramData\\SF\\DiagnosticsStore"

    After this, I deinstalled "Microsoft Azure Service Fabric" from all three machines using the "Programs & Features"-GUI under Win Server 2016 and rebooted each machine.

    After the reboots I just ran the exact same CreateServiceFabricCluster as the first time, only using the updated Cluster Config.

    Things went well, no errors in the Service Fabric Explorer after the reinstall.

    Testing the Start-ServiceFabricClusterConfigurationUpgrade also went well, upgrading eacht of my configured failure domains one after another. Until the complete reinstall of SF I got the reported Error: "The first Fabric upgrade must specify both the code and config versions"

    Last But not Least, my Cluster-Details Page in the Service Fabric Explorer now doesn't show ConfigVersion = 0 anymore, but ConfigVersion = 1.

    I don't know if there would have been easier solutions than a complete reinstall. But it did the trick for me so I tought about sharing my expiriences with you.

    Good Luck solving the problem on your side!

    Michael



    Friday, June 16, 2017 11:53 AM