Skip to content

Frequently asked questions

Project management

Listing all Virtual Machines in all projects

If you are member or owner of multiple project, it may be challenging to match particular Virtual Machines to their hosting project. To list all your projects, and the hosted VMs, you can run:

$ openstack project list -f value -c Name | while read -r project; do echo $project ; OS_PROJECT_NAME="$project" openstack server list ; done

General

Not keeping time in guests

The guest VMs will occasionally time clocks will occasionally drift. To ensure time is kept in sync, the CERN time server can be used. For Puppet managed machines, NTP is configured by default.

For Linux hosts which are manually managed, you can use ntpd and ntpdate to synchronise the host.

$ sudo yum install -y ntp
$ sudo ntpdate ip-time-1

Linux

Graphical interface on Linux VMs

Recipes to install a graphical interface exist here:

Can I extend a volume or shall I create a new one and synchronize the data?

First of all, volumes can be extended live if you have enough quota on the project, with no downtime.

Volume snapshots in our infrastructure are just pointers in the RW history of the volume. It takes a few seconds to take it. You can do it even when the volume is in-use, it will create this pointer before any subsequent write.

Once you have an snapshot, you can create as many volumes as you want based on this one and make them even with a bigger size. Every single of them will contain exactly the same data as the original. The time to create this volume depends. For example if you create a volume with a different type, from standard to cp1, it may need to copy data accross (different power zones)

Just to let you know, once you create a volume snapshot, it is linked to the volume history, so you may not be able to delete the original volume until all his snapshots are deleted.

Both approaches are valid, it really depends on your application, the amount of data that changes. You can even merge both to do the first copy of data with snapshots, and use rsync for the final copy at the moment of the intervention

I am unable to delete a volume, the operation returns BadRequest(400)

If you delete a volume and the api returns something like this:

BadRequest: Failed to delete volume with name or ID: Invalid volume: Volume status must be available or error or error_restoring or error_extending or error_managing and must not be migrating, attached, belong to a group, have snapshots or be disassociated from snapshots after volume transfer. (HTTP 400)
Please check if the volume has any volume snapshots. You need to delete them first, before triggering the deletion of the volume

I am unable to get IPv6 connectivity on the Prevessin Data Centre

The OpenStack deployment in the Prevessin Data Centre does not support the Privacy Extensions defined by RFC 4941, or the Opaque Identifier generation methods defined in RFC 7217 as explained in the following link

The interface identifier and DUID used must be directly derived from the MAC address as described in RFC 2373. The virtual machines must not be set up to utilize either of these methods when generating their interface identifier, or they might not be able to communicate properly on the network. For example, in Linux guests, these are controlled via these two sysctl variables:

net.ipv6.conf..use_tempaddr (Privacy Extensions) net.ipv6.conf..addr_gen_mode

Both values must be set to zero to ensure IPv6 connectivity of the virtual machines.

Alternatively with the networkmanager config:

# /etc/NetworkManager/conf.d/90-dhcp.conf

[connection]
ipv6.addr-gen-mode=0

I am requested to recreate my virtual machine but none of the available flavors have the same disk space

The flavors available in the OpenStack cloud match the ratios in terms of CPUs, RAM or Disk of the servers that are running them. Sometimes they are limited due to service operability. We keep the disk size controlled to maximize uptime in case of a HW event, as we always evacuate the VMs hosted without stopping them, to minimize impact on the end users.

These hardware constraints are exposed to end-users through flavor families. When we need to replace hardware with a certain ratio as it gets decommissioned and the new hardware has a much higher ratio, we may ask users to recreate their virtual machines if they do not fit on the replacement hardware.

Please follow this decision tree

graph TD
A[Recreate VM] --> B{Do I really need the disk space?}
    B -->|No| C[Use an existing flavor with same number of cores or memory]
    B -->|Yes| D{Can I use a volume instead?}
    D -->|Yes| E[Create a VM with an existing flavor and attach an additional volume]
    D -->|No| F[Open a ticket to the Cloud Infrastructure Service for support]