IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
This commits introduces a new framework for OpenNebula, it is based on
standard the OpenNebula middleware access driver (mad) model. Benefits
compared to the current implementation:
* Less overhead on API server and streamlined communications
* Ease integration with third party schedulers
* New "optimize" action to balance the workload of a cluster
* Embedded Plan Manager to apply optimization and deployment plans with
custom parameters for the scheduling window
* Custom optimization policies per cluster
This commits introduces 2 schedulers:
* Rank scheduler based on the match-making algorithm present in 6.x
series
* OpenNebula DRS an optimization algorithm based on an ILP solver for
placement and cluster optimization
Also new API calls has been added for onecluster as well as the CLI
updates to expose them.
co-authored-by: MarioRobres <mrobres@opennebula.io>
co-authored-by: Pavel Czerny <pczerny@opennebula.io>
co-authored-by: Mirko Stojiljkovic <mstojiljkovic@opennebula.io>
co-authored-by: Valentyn Bohdan <vbohdan@opennebula.io>
This feature auto mounts NFS volumes for datastores. The behavior can be controlled with the following template attributes:
- ANFS_ENABLE: YES/NO to enable the automatic NFS management
- ANFS_HOST: hostname/IP of the NFS server
- ANFS_PATH: of the NFS share for this datastore
- ANFS_OPTS: Comma separated options (fstab-like) used for mounting the NFS shares
The shared drivers also performs a "garbage-collector" like to unmount NFS volumes on the hypervisors when the datastores are deleted. This operation is performed in a "lazy" way, when VMs are deployed on the hypervisor.
* oneimage create: mount/umount DS on demand
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* Move sh mount/umount logic to libfs.sh
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* oneimage delete: mount/umount DS on demand
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* Limit NFS mount/umount from SH code to 30s
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* oneimage clone: mount/umount DS on demand
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* onevm deploy: mount DS on demand
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* Temporal mount only when needed
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* onevm deploy: mount DS on demand (persisten)
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* Linter
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* onevm deploy: umount unneeded DS
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* Allow mount/umount/sed from sudoers
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* Temporal mount: use NFS_AUTO_OPTS
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* Do AutoNFS cleanup even during non-related ops
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* Final refactor and improvements
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* ro values for Datastore object
* Wording for AutoNFS RO attributes
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* NFS_AUTO_ENABLE: change 'true' to 'yes'
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* sudoers: merge everythng into ONE_NFS
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* Quote NFS_AUTO_OPTS
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* Fix difference between sudoers specs
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* Touch .automounted in AutoNFS mounts
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
* Update sudoers.rb
* Update rm
* Quote bash variable expansion
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
---------
Signed-off-by: Guillermo Ramos <gramos@opennebula.io>
Co-authored-by: Ruben S. Montero <rsmontero@opennebula.org>
This feature adds support for VM quotas at cluster level. CLUSTER_IDS
can be set for a VM quota so the admin can limit the number of resources (e.g. VMs) a
user/group can create in a given cluster.
This commit also fixes:
- B #2226
- B #6823
Co-authored-by: Pavel Czerny <pczerny@opennebula.io>
This commit adds the following features using the vlan_filter
capabilities of the Linux Bridge:
* 802.1Q driver, support for QinQ. This is not a full implementation
compared to the vSwitch as there is no native support in the linux
bridge:
- VLAN_ID is the (outer/C-VLAN) tag implemented by tagging the
PHY_DEV interface
- CVLANS are the customer tags, any other VLAN is filtered out in
the bridge
* bridge driver, support for vlan filtering (VLAN_TAGGED_ID) only the
specified tags are allowed in the trunk.
Additionally this commits includes linting of some files.
(cherry picked from commit e379bd66842dc4aa256e1c503ca425553d02bf9b)
This is the first version of a ScaleWay drivers for OneProvision. The
drivers include the following features:
- Full support for defining Scaleway zones and regions, predefined files
for Paris-1, Amsterdam-1 and Warsaw-3.
- Edge cluster based on metal instances
- Flexible IP support for VMs running on Scaleway clusters
- Private networking based on VXLAN and FRR (bgp-evpn). Future versions
of this driver will consider extending IPAM to use Scaleway private
networks.
- Playbook includes an /etc/hosts update to support name resolution of
scaleway instances (needed by libvirt for live migration)
This commit also updates firewall to allow libvirt ports.
co-authored-by: Victor Palma <vpalma@opennebula.io>
- Now are named local. SSH drivers will be distributed for running VMs
that use the SSH drivers.
- The new drivers optimize Qcow2 operations in the same way as the
shared TM driver, including:
- thin provisioning, when available a cow copy is made
- Qcow2 snapshots for the snapshot operations
- Some operations has been improved code-wise
- It should be a drop-in replacement for ssh
- New drivers are written in Ruby to accommodate future improvements
- By default new datastores will use "local" instead of "ssh"
co-authored-by: Ruben S. Montero <rsmontero@opennebula.org>
* Remove vm_import table from DB
* Remove imported vms actions
* Fix fsck for image and network
* onedb fsck fix running_vms only for non-backup images
This feature let's cloud admin to proxy VM connections to any service through the hypervisor. VMs uses a link local IP that is forwarded to a local proxy. This simplifies VM network requirements, e.g. to access gateway, vaults, configuration services etc...
Implementation:
- Use network namespaces to isolate VNET networking. ip netns command is executed through a wrapper to limit sudo access to commands.
- Add tproxy.rb app to manage a group of daemons on HV nodes.
- Use unix sockets for communication between proxy peers. "Inner" proxy runs in the netns without any network access. "Outer" proxy handles HV connections to services.
- Use OpenNebulaNetwork.conf + 'onehost sync -f' for configuration. Proxy can be defined per network.
* Cleanup oned in case of initialization error
* Fix monitoring after onemonitord restart. The code includes a "hook" point in case a driver is re-started so custom code can be executed. InformationManager sends the list of hosts and raft status in this case.
* B #5801: Update error msg, in case of duplicated drivers
