Server management for SDL Content Manager

I'd appreciate it if anyone give me suggestion regarding Windows update management for servers of SDL Content Manager.

In our company, all Windows servers must be installed Windows update using WSUS (Windows Server Update Services) once a month. In general, the servers must be restarted after the Windows update installation. The servers consist of three application servers and a database server. When rebooting these servers simultaneously, the application server fails to connect the database. This results some Content Manager services fail to start. Is there any method to avoid such a trouble? I heard from SDL support that it's possible to increase the maximum retry count for the database connection by modifying the registory. But it's not sophisticated way.

Best regards,

Naoki

Parents
  • Hi Naoki,

    As you noticed there is a sequence and dependency on how software connects and re-connects. Key question often is how often and how long should those reconnects happen. In the end software doesn't know if the database server disappeared (disk crash, new name, etc) or if it is just a network glitch expressed in seconds. When should the client of the database push the alarm button that a critical part of the system (the database) is missing.

    The above is somewhat different every time, definitely if it are unexpected outages. In you case however you describe a controlled outage, it happens regularly and predictably.

    So my first suggestion would be to have updates on the database server first, then the application servers (and typically the database server is already back when the application servers come back from a reboot or so). Can you control the sequence and the duration of the updates?

    Even cleaner would be to do a clean stop of the software on the application servers, do all updates on all servers, and then do a clean start. For 'SDL Tridion Docs Content Manager' we offer a Windows PowerShell module called ISHDeploy. Among many other cmdlets there are some cmdlets worth mentioning here: 

    1. Stop-ISHDeployment which knows per Content Manager version which services should be stopped nicely and in what order
    2. Start-ISHDeployment which knows per Content Manager version which services should be started nicely and in the right order
    3. Restart-ISHDeployment in essence runs Stop-... and Start-... back to back, so it restarts Content Manager services without a reboot

    The PowerShell module ISHDeploy is on PSGallery just like ISHRemote. Handy resources are

    1. https://sdl.github.io/ISHDeploy/ on some getting started material depending on your use case
    2. https://vimeo.com/378093731 is a video where I talk you through some automation concept

    Best wishes,
    Dave

Reply
  • Hi Naoki,

    As you noticed there is a sequence and dependency on how software connects and re-connects. Key question often is how often and how long should those reconnects happen. In the end software doesn't know if the database server disappeared (disk crash, new name, etc) or if it is just a network glitch expressed in seconds. When should the client of the database push the alarm button that a critical part of the system (the database) is missing.

    The above is somewhat different every time, definitely if it are unexpected outages. In you case however you describe a controlled outage, it happens regularly and predictably.

    So my first suggestion would be to have updates on the database server first, then the application servers (and typically the database server is already back when the application servers come back from a reboot or so). Can you control the sequence and the duration of the updates?

    Even cleaner would be to do a clean stop of the software on the application servers, do all updates on all servers, and then do a clean start. For 'SDL Tridion Docs Content Manager' we offer a Windows PowerShell module called ISHDeploy. Among many other cmdlets there are some cmdlets worth mentioning here: 

    1. Stop-ISHDeployment which knows per Content Manager version which services should be stopped nicely and in what order
    2. Start-ISHDeployment which knows per Content Manager version which services should be started nicely and in the right order
    3. Restart-ISHDeployment in essence runs Stop-... and Start-... back to back, so it restarts Content Manager services without a reboot

    The PowerShell module ISHDeploy is on PSGallery just like ISHRemote. Handy resources are

    1. https://sdl.github.io/ISHDeploy/ on some getting started material depending on your use case
    2. https://vimeo.com/378093731 is a video where I talk you through some automation concept

    Best wishes,
    Dave

Children