cloudstack-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Yiping Zhang <>
Subject Re: R: A Story of a Failed XenServer Upgrade
Date Fri, 08 Jan 2016 21:52:18 GMT
Since I can’t use attachment,  I just include the doc in the message. Hopefully, the indentations
and formats will go through properly.


XenServer pool manual upgrade from 6.2 to 6.5 using ISO
Reference article for upgrading XenServer pool used for Cloudstack

Manual upgrade to XenServer from 6.2 to 6.5 using ISO

On CloudStack Management server

  *   Edit the file /etc/cloudstack/managment/ to include the following
line at the end:
     *   manage.xenserver.pool.master=false
  *   Restart cloudstack-management service
     *   service cloudstack-management restart

Pre-upgrade steps

  *   Disable XenServer pool HA from XenCenter or CLI
  *   Backup XenServer resource pool configurations
     *   Take screen shot for pool network settings in XenCenter
     *   Take notes of Storage Repo mount points and NFS volumes.

Inside CloudStack Web UI

  *   Put pool master host into Maintenance (CLOUDSTACK ONLY!). This should migrate all VM
instances currently running on the pool master onto other hosts
  *   Unmanage the cluster.  This should make the hypervisors show as disconnected in the
UI.  PLEASE MAKE SURE THAT YOU CLICK "Unmanage Cluster", NOT "Disable Cluster"!!!

On Pool Master

  *   Connect to physical console using DRAC/iLO/equivalent
  *   Attach XenServer 6.5 ISO as virtual DVD
  *   Verify that this host uses Legacy BIOS rather than UEFI, as UEFI is NOT supported by
XenServer.  (May not be needed, as XS 6.2 also requires Legacy BIOS)
  *   Reboot physical server
  *   Once the server boots up off DVD image:
Note: we used an answer file to allow automated upgrade.  You can just manually do the upgrade
     *   At the first prompt, hit F2 to get the advanced menu, which won't time out quickly
     *   Type:
        *   menu.c32
     *   Hit Enter, then hit the Tab key.
     *   You will be presented with a boot line that looks similar to the following:
        *   mboot.c32 /boot/xen.gz dom0_max_vcpus=2 dom0_mem=752M com1=115200,8n1 console=com1,vga
— /boot/vmlinuz xencons=hvc console=hvc0 console=tty0 — /install.img
     *   You will need to edit this line to include the bold parts.  Nothing else in this
line needs to be changed:
        *   mboot.c32 /boot/xen.gz dom0_max_vcpus=2 dom0_mem=752M com1=115200,8n1 console=com1,vga
— /boot/vmlinuz xencons=hvc console=hvc0 console=tty0 answerfile=http://server_ip/path/to/answer-file.xml
— /install.img
     *   Hit Enter and watch the system upgrade itself.
     *   Once complete, eject the DVD image and reboot the host.

Verify network and storage settings of upgraded host (in XenCenter):

  *   Configure networks if necessary, just in case any additional NIC's need to be configured
  *   Repair the HA SR, if necessary

Inside CloudStack Web UI

  *   Re-manage the cluster
  *   Wait for all hosts to be in the Up state (except the pool master, which will stay as
  *   Wait for all SR are connected and online
  *   Take the pool master out of Maintenance mode

On the CloudStack management server in a terminal window

  *   Undo the change in /etc/cloudstack/management/ file
  *   Restart cloudstack-management service and wait for the dust to settle on all servers/storage
being visible. All hosts should be in Up state, including the pool master.

Now for each slave hosts:

    In CloudStack web UI

  *   Put slave host into Maintenance mode (CLOUDSTACK ONLY!), in order to evacuate all instances
running on this host
  *   On physical console, follow the same steps to attach DVD image and do upgrade as for
pool master node
  *   Perform the eject, reboot, and and host verification as listed above
  *   Verify the networks and that it has properly rejoined the Xenserver Pool (in XenCenter)
  *   Take the host out of Maintenance mode in the CloudStack UI

Post-operation steps for each hosts:

  *   Confirm XenCenter Licensing
     *   You may need to remove the license from the cluster before applying a new one
  *   Enable HA

You are done !
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message