Relocation of central IT infrastructure of the IT Center Q1 2016

Partial interruption of the IT Center services, in particular of the services Phsyical Server Hosting and Virtual Server Hosting.

What happened?

Part of the virtual machines (Virtual Server Hosting) and physical servers (Physical Server Hosting) hosted by the IT Center of the RWTH have to be shut down for several hours for the purposes of a relocation and a maintenance announced for 23 and 24 January 2016. If one or several of your virtual machines or physical servers are affected by this shutdown, the IT Center will send you an e-mail in December to personally inform you accordingly. For details and news, please refer to the fault report pages of the IT Center at the following address:

http://www.itc.rwth-aachen.de/cms/IT-Center/Dienste/Status-und-Stoerungsmeldung/~fnum/Stoerungsmeldungen/

Why?

As a result of the preparatory measures for the establishment of the next generation of the RWTH Aachen high performance computer, extensive construction measures are required at the IT Center in collaboration with the BLB NRW (Building and Property Management NRW).
These particularly serve to make operation of the computer infrastructure more energy efficient in the future. In order to prepare these on schedule, it is necessary to clear a machine hall of the IT Center at the location Seffenter Weg 23 for further steps. Unfortunately, due to the closely interconnected system infrastructures, this is not possible using the usually available transparency and redundancy properties of the services considering the time requirements.

What is the procedure?

Maintenance for the relocation will take place in the following steps:

  1. At the beginning of the maintenance, the operating status (Powered On/Supsended oder Powered Off) of the affected VMs and physical servers are determined and saved. The aim is to ensure that after the maintenance the previous state of the VMs and servers can be restored.   
  2. Automatic shutdown of all remaining servers
    1. Concerning virtual machines (VMs):
      A shutdown signal will be sent to all VMs that do not have the status "Powered Off". In order for this signal to result in a correct shutdown of the VMs, it is necessary that the "VMware Tools" or the "Open VM Tools" are installed and running in an updated version. It is recommended to test this feature in advance by selecting the feature "Power" --> "Shutdown Guest" (the VM will be shut down) at the vSphere Client for the affected VMs. Information on accessing your VMs can be found here: https://doc.itc.rwth-aachen.de/display/VHOST/Nutzung
    2. Concerning physical servers:
      Servers will be sent a ACPI-signal for the shutdown of the systems. This signal usually correspondsto a quick push on the power button of the server. It is recommended to test the feature beforehand from the LOM of the server by going to "Power Management" --> "Server Power" and inducing a "Momentary Press" (The procedure can vary due to different server models. Information on accessing your physical servers can be found here: https://doc.itc.rwth-aachen.de/display/PHOST/Nutzung
    3. If the VMs or physical servers are not shut down within 10 minutes after the shutdown signal was sent, the shutdown will be forced. This corresponds to pulling out the power plug on a physical computer. In order to prevent delays e.g. due to updating processes, we recommend to test the shutdown or restart of your VM or your servers a few days before the maintenance appointment.
  3. Afterwards, the actual maintenance takes place by relocating the IT infrastructure.
  4. On completion of the maintenance, the VMs and physical servers are rstored to the operational state which they were in at the beginning. Please note that the restart of the VM or physical servers happens with a time lag and can take some time due to e.g. Filesystem Check Processes.
    Please also note that you may have to conduct manual interventions for the restoration of your server services.

Should your VM or physical server be incompatible with the process described above (e.g. due to applications that have to be stopped manually), we recommend to independently shut down the VMs or physical servers prior to the commencement of the maintenance period. In this case, you will also have to restart them after the maintenance. Please do not fail to note the reports on the fault report page of the IT Center and start your VMs or phyiscal servers only when the maintenance was explicitly reported as completed.

  • No labels