LXD 4.13 has been released

Introduction

The LXD team is very excited to announce the release of LXD 4.13!

This comes with a lot of exciting new features, including project resource usage, NVIDIA MIG support, snapshot schedule aliases, default cluster architecture and a variety of improvements to the CLI tool.

Enjoy!

New features and highlights

Support for instance filters in lxc list

New filtering options are now available in lxc list as key=value filters.

  • type= (container or virtual-machine)
  • status= (running, frozen, stopped or broken)
  • architecture= (a valid architecture name)
  • location= (a cluster server name)
  • ipv4= (specific address or CIDR)
  • ipv6= (specific address or CIDR)

Example with IPv6 CIDR filtering:

stgraber@penguin:~$ lxc list ipv6=2001:470:b368::/48
+----------+---------+-----------------------+----------------------------------------------+-----------------+-----------+----------+
|   NAME   |  STATE  |         IPV4          |                     IPV6                     |      TYPE       | SNAPSHOTS | LOCATION |
+----------+---------+-----------------------+----------------------------------------------+-----------------+-----------+----------+
| buildd01 | RUNNING | 10.232.117.1 (lxdbr0) | 2602:fd23:8:1005:216:3eff:fe19:fd6e (enp5s0) | VIRTUAL-MACHINE | 7         | abydos   |
|          |         |                       | 2001:470:b368:4242::1 (lxdbr0)               |                 |           |          |
+----------+---------+-----------------------+----------------------------------------------+-----------------+-----------+----------+
| buildd02 | RUNNING | 10.181.173.1 (lxdbr0) | 2602:fd23:8:1005:216:3eff:fec8:841b (enp5s0) | VIRTUAL-MACHINE | 8         | langara  |
|          |         |                       | 2001:470:b368:4242::1 (lxdbr0)               |                 |           |          |
+----------+---------+-----------------------+----------------------------------------------+-----------------+-----------+----------+
| buildd03 | RUNNING | 10.232.46.1 (lxdbr0)  | 2602:fd23:8:1005:216:3eff:fe67:f919 (enp5s0) | VIRTUAL-MACHINE | 7         | orilla   |
|          |         |                       | 2001:470:b368:4242::1 (lxdbr0)               |                 |           |          |
+----------+---------+-----------------------+----------------------------------------------+-----------------+-----------+----------+

NVIDIA MIG support for containers

A new GPU type was introduced to support NVIDIA Multi-Instance GPUs.
On systems supporting the feature, it’s now possible to pass pre-created MIG compute instances into LXD containers.

This is done using a new mig gputype for the gpu device type.

lxc config device add c1 gpu0 gpu gputype=mig mig.gi=5 mig.ci=1 pci=09:00.0

This will use the compute instance 1 of the GPU instance 5 from the GPU at address 09:00.0.
You can find all the relevant values from nvidia-smi after having created your MIG instances.

System wide remote configuration

The lxc command line tool now considers /etc/lxd/ as a source of configuration.
A config.yaml placed there will be merged with the user’s own configuration.

This allows for system-wide configuration of LXD remotes. Server certificates can similarly be placed in /etc/lxd/servercerts/.

NOTE: This won’t work for snap users yet, we will soon be setting things up there so the global configuration can be placed inside /var/snap/lxd/common/.

Project resource usage

For projects with resource limits in place, it’s now possible to get a simple overview of the limits and current usage. Note that the usage will only be accurate if a limit is set.

stgraber@penguin:~$ lxc project info nsec
+------------------+-----------+---------+
|     RESOURCE     |   LIMIT   |  USAGE  |
+------------------+-----------+---------+
| CONTAINERS       | UNLIMITED | 3       |
+------------------+-----------+---------+
| CPU              | 4         | 3       |
+------------------+-----------+---------+
| DISK             | 53.69GB   | 33.29GB |
+------------------+-----------+---------+
| INSTANCES        | UNLIMITED | 3       |
+------------------+-----------+---------+
| MEMORY           | 4.29GB    | 3.22GB  |
+------------------+-----------+---------+
| NETWORKS         | 10        | 1       |
+------------------+-----------+---------+
| PROCESSES        | 100000    | 30000   |
+------------------+-----------+---------+
| VIRTUAL-MACHINES | UNLIMITED | 0       |
+------------------+-----------+---------+

Snapshot schedule aliases

LXD supports automatically scheduling snapshots for both instances and for custom storage volumes. This is done using the snapshots.schedule configuration option which takes a cron pattern.

Now on top of those cron patterns, the following aliases are also supported:

  • @hourly
  • @daily
  • @midnight
  • @weekly
  • @monthly
  • @annually
  • @yearly

And for instances only, there’s an additional @startup which can be used to snapshot on every instance startup or restart.

The other advantage of those aliases over a matching cron pattern is that those will be randomly distributed over time. So a @daily will not trigger at the same time for all instances or all storage volumes but it will trigger at the same time for that one instance or storage volume.

images.default_architecture for multi-architecture setups

For those running multi-architecture clusters, launching an instance without a specific target and using an image which is available on multiple architecture would result in LXD finding the least busy of all servers regardless of architecture and creating the instance there.

That’s rarely the desired behavior as most multi-architecture cluster have a primary architecture (often x86_64) and then a bit of capacity on less common architecture (aarch64, ppc64le, s390x, …). So LXD treating them all as equal and just spreading the instances among them may be quite confusing.

The new configuration option, images.default_architecture can be set to an architecture name which will be used for any cases where the image isn’t specific to a single architecture.

For example on a multi-architecture cluster:

lxc config set images.default_architecture x86_64
lxc launch images:ubuntu/20.04 c1

Will now always deploy as x86_64 and one will have to use:

lxc launch images:ubuntu/20.04/arm64 c1

To specifically request the aarch64 image which will then pick a cluster server capable of running aarch64 images.

New description column in lxc project list, lxc profile list and lxc storage list

With this change just about every top level list command now shows the description attribute. This is now consistently available on:

  • lxc profile list
  • lxc image list
  • lxc project list
  • lxc storage list
  • lxc network list

It’s just not shown by default on lxc list due to lack of horizontal space there but it’s one of the configurable columns so can be enabled there too.

Reworked handling of default action in network ACLs

The recently introduced network ACLs came with default.action and default.logged config options at the ACL level. This was a bad design as many ACLs can be applied to one instance making those config keys confusing.

As a result, we made the decision to completely drop those two configuration keys from the ACL configuration and instead introduce a bunch more as direct configuration options on instance network devices and networks.

New configuration keys:

  • security.acls.default.egress.action
  • security.acls.default.egress.logged
  • security.acls.default.ingress.action
  • security.acls.default.ingress.logged

This also separates ingress and egress making it easy to log any blocked egress traffic while keeping blocked ingress traffic silently ignored.

--console with lxc stop

lxc stop now joins lxc start and lxc restart in supporting a --console option.

With it, you can attach to the instance’s console just as it gets signaled to shutdown.
This is useful to observe the shutdown sequence and spot any error.

More auto-generated REST-API documentation

Everything except for /1.0/instances and /1.0/storage-pools is now covered in our Swagger API files. This should cover all the different parameters and ways to interact with those LXD APIs.

A temporary rendering of this can be found here: https://dl.stgraber.org/swag-lxd/

Complete changelog

Here is a complete list of all changes in this release:

Full commit list
  • lxd/migration: Refresh protobuf
  • lxd/storage/ceph: Fix ceph.rbd.features
  • lxd/cluster/gateway: Expose node is not clustered error
  • lxd/cluster/gateway/test: TestGateway_Single ErrNodeIsNotClustered usage
  • lxd/patches: Adds db_nodes_autoinc patch
  • lxd/storage/drivers/driver/ceph/utils: util.SplitNTrimSpace usage
  • lxd/network/openvswitch/ovs: Adds TCP flag constants
  • lxd/network/acl/acl/ovn: Switches to openvswitch TCP flag constants in OVNApplyNetworkBaselineRules
  • shared/util: Removes GetStableRandomInt64FromList and GetStableRandomGenerator from shared pkg
  • lxd/util/random: Adds GetStableRandomGenerator and GetStableRandomInt64FromList functions
  • lxd/images: util.GetStableRandomInt64FromList usage
  • lxd/network/driver/bridge: util.GetStableRandomGenerator usage
  • lxd/network/driver/ovn: util.GetStableRandomGenerator usage
  • lxd/storage/volumes/snapshot: util.GetStableRandomInt64FromList usage
  • test: Run test_clustering_image_refresh
  • shared/api/netork/acl: Adds missing example doc fields
  • lxc/network/acl: Update field examples
  • i18n: Update translation template
  • lxd/network/acl/acl/load: Fix UsedBy with project profiles
  • test/suites: Fix sed command
  • shared/api: Mark most ACL rule fields omitempty
  • Fix typo in doc/projects.md, replace images with backups
  • Fix a typo in rest-api.md for renaming a network ACL
  • client/lxd: Don’t treat % chars from LXD server response as placeholders in lxdParseResponse
  • doc/network/acls: Changes reserved subject names
  • lxd/network/acl/acl/ovn: Handles deprecated aliases for subjects
  • shared/api/network/acl: Updates example with non-depcared reserved subject
  • lxd/network/acl/driver/common: Deprecates #internal and #external subjects and replaces them with @internal and @external
  • Doc: fix typo on network type
  • doc/rest-api: Refresh swagger YAML
  • doc/rest-api: Refresh swagger YAML
  • lxd: Unmount image and backup volume on shutdown
  • lxd: Fix static analysis
  • shared: Remove WriteTempFile
  • shared/osarch: Remove dependency on shared package
  • shared/osarch: Add SupportedArchitectures
  • shared/validate: Add IsArchitecture
  • lxd/project: Add images.default_architecture
  • lxd/cluster: Add images.default_architecture
  • lxd/db: Support images.default_architecture
  • lxd: Support images.default_architecture
  • doc: Add images.default_architecture
  • api: images_default_architecture
  • lxd/network/driver/ovn: Adds n.ensureNetworkPortGroup and use it in setup() and Start()
  • lxd/network/driver/ovn: Clarifies comment
  • lxd/storage/drivers/generic/vfs: Error quoting in genericVFSVolumeSnapshots
  • lxd/storage/drivers/driver/btrfs/volumes: Ensure subvolumes in snapshots are copied during CreateVolumeFromCopy
  • lxd/images: Optimize image distribution
  • lxd: Properly unmount old image volume
  • lxd/network/acl/acl/ovn: Ensure to parenthesised || when used with && in rule match
  • lxd/instance/qemu: Add 5s QMP timeout
  • lxd/storage/drivers/volume: Optimized creation of slice in Snapshots()
  • lxd/storage/drivers/volume: Adds SnapshotsMatch function
  • lxd/storage/drivers/interface: Updates BackupVolume to take a slice of snapshot names
  • lxd/db/instances: Clarifies expected return order of snapshots in GetInstanceSnapshotsWithName
  • lxd/storage/backend/lxd: Updates BackupInstance to provide list of snapshot names to b.driver.BackupVolume
  • lxd/storage/backend/lxd: Updates BackupCustomVolume to provide list of snapshot names to b.driver.BackupVolume
  • lxd/storage/drivers/generic/vfs: Updates genericVFSBackupVolume to accept a list of snapshot names
  • lxd/storage/drivers/driver/btrfs/volumes: Updates BackupVolume to accept a list of snapshot names
  • lxd/storage/drivers/driver/ceph/volumes: Updates BackupVolume to accept a list of snapshot names
  • lxd/storage/drivers/driver/cephfs/volumes: Updates BackupVolume to accept a list of snapshot names
  • lxd/storage/drivers/driver/dir/volumes: Updates BackupVolume to accept a list of snapshot names
  • lxd/storage/drivers/driver/lvm/volumes: Updates BackupVolume to accept a list of snapshot names
  • lxd/storage/drivers/driver/mock: Updates BackupVolume to accept a list of snapshot names
  • lxd/storage/drivers/driver/zfs/volumes: Updates BackupVolume to accept a list of snapshot names
  • lxd/storage/drivers: Clarifies ordering of VolumeSnapshots and volume.Snapshots() in comment
  • lxd/storage/volumes/backup: Use requested project name for operation
  • lxd/storage/volumes/snapshot: Use requested project name for operation
  • lxd/storage/volumes: Use requested project name for operation
  • lxd/operations: Renames project to projectName
  • lxd: Renames project to projectName
  • clustering: Force config when bootstrapping
  • lxd/network/openvswitch/ovn: Work around ovn-nbctl NAT bugs in LogicalRouterDNATSNATAdd
  • lxd/network/openvswitch/ovn: Use same technique as LogicalRouterDNATSNATAdd when mayExist is true
  • lxd/network/openvswitch/ovn: Merge LogicalSwitchPortAdd and LogicalSwitchPortSet
  • lxd/network/driver/ovn: client.LogicalSwitchPortAdd usage
  • lxd/network/openvswitch/ovn: Reduce ovn-nbctl calls in LogicalSwitchPortDeleteDNS
  • lxd/network/openvswitch/ovn: Reduce calls to ovn-nbctl in LogicalSwitchPortLinkRouter
  • lxd/network/openvswitch/ovn: Reduce calls to ovn-nbctl in LogicalSwitchPortLinkProviderNetwork
  • lxd/network/openvswitch/ovn: Updates LogicalRouterDNATSNATDelete to support removing multiple entries in single call
  • lxd/network/driver/ovn: Updates InstanceDevicePortDelete to always attempt to remove its IPs from DNAT entries
  • lxd/network/openvswitch/ovn: Updates LogicalRouterRouteDelete to support removing multiple routes
  • lxd/network/driver/ovn: LogicalRouterRouteDelete usage to reduce ovn-nbctl calls
  • lxc/network_acl: Fix example
  • i18n: Update translation templates
  • lxc/project: Show description
  • lxc/profile: Add description column
  • lxc/storage: Add description column
  • doc: improve wording of network ACLs
  • doc/instances: Adds ipv4.routes and ipv6.routes for OVN NICs
  • test/main: Fix clustering test typo
  • test/clustering: Cleanup properly
  • test/clustering: s/localhost/public/
  • test/clustering: Replace sleep with wait
  • test/clustring: Unset LXD_NETNS
  • lxd/device/disk: Fallback to using mount device path for major/minor number extraction for BTRFS
  • lxd/instance/drivers/driver/lxc: Ensure container is stopped if post start hooks fail
  • lxd/instance/drivers/driver/qemu: Adds comment clarifying order of post start hook failure actions
  • lxd/network/openvswitch/ovn: Adds LogicalSwitchPorts function
  • lxd/network/driver/ovn: Use client.LogicalSwitchPorts in handleDependencyChange and Update
  • tests: Don’t block on /dev/random
  • lxd/network/acl/acl/ovn: Removes unused state arg from ovnApplyToPortGroup
  • lxd/network/driver/ovn: Improved comment in InstanceDevicePortAdd
  • lxd/db/images: Fix incorrect cached attribute handling
  • lxd/images: Improve error logging in autoUpdateImages
  • doc: Added a Table of Content in /doc/index.md
  • doc: Added “Feature Requests” to Bug Reports section in /doc/index.md
  • lxd/patches: Fix typo in patchNetworkClearBridgeVolatileHwaddr
  • lxd/patches: Adds patchNetworkACLRemoveDefaults to remove ACL default.action and default.logged keys
  • doc/network-acls: Removes references to default.action and default.logged
  • shared/api/network/acl: Removes reference to default.action in config
  • lxd/network/acl/acl/ovn: Removes default.action and default.logged behavior
  • lxd/network/acl/driver/common: Removes references to default.action and default.logged in config validation
  • doc/rest-api: Refresh swagger YAML
  • lxd/network/driver/ovn: Renames instance device structs to OVNInstanceNICSetupOpts and OVNInstanceNICStopOpts
  • lxd/network/driver/ovn: Renames InstanceDevicePortConfigParse to instanceDevicePortRoutesParse
  • lxd/network/driver/ovn: Renames InstanceDevicePortAdd to InstanceDevicePortSetup
  • lxd/network/driver/ovn: Updates InstanceDevicePortDelete to accept OVNInstanceNICStopOpts
  • lxd/network/driver/ovn: Updates handleDependencyChange to use n.InstanceDevicePortSetup
  • lxd/device/nic/ovn: Updates ovnNet interface
  • lxd/device/nic/ovn: d.network.InstanceDevicePortSetup and d.network.InstanceDevicePortDelete usage in Start()
  • lxd/device/nic/ovn: d.network.InstanceDevicePortSetup usage in Update()
  • lxd/device/nic/ovn: d.network.InstanceDevicePortDelete usage in Stop()
  • lxd/network/acl/driver/common: Export ValidActions for network and NIC validation
  • lxd/device/nic: Add support for validating security.acls.default.{in,e}gress.{action,logged} config keys
  • lxd/device/nic/ovn: Validates security.acls.default.{in,e}gress.{action,logged} config keys
  • doc/networks: Adds security.acls.default.{in,e}gress.{action,logged} keys to ovn network
  • doc/instances: Adds security.acls.default.{in,e}gress.{action,logged} keys to OVN NIC
  • api: Adds network_ovn_acl_defaults extension
  • lxd/network/openvswitch/ovn: space trimming cleanup
  • lxd/network/openvswitch/ovn: Renames setACLRules to aclRuleAddAppendArgs
  • lxd/network/openvswitch/ovn: Updates LogicalSwitchSetACLRules to use o.aclRuleAddAppendArgs
  • lxd/network/openvswitch/ovn: Updates PortGroupSetACLRules to use o.aclRuleAddAppendArgs
  • lxd/network/openvswitch/ovn: Adds logicalSwitchPortACLRules function
  • lxd/network/openvswitch/ovn: Updates LogicalSwitchPortDeleteDNS to use o.logicalSwitchPortDeleteDNSAppendArgs
  • lxd/network/openvswitch/ovn: Adds logicalSwitchPortDeleteAppendArgs function
  • lxd/network/openvswitch/ovn: Updates LogicalSwitchPortDelete to use o.logicalSwitchPortDeleteAppendArgs
  • lxd/network/openvswitch/ovn: Adds LogicalSwitchPortCleanup function
  • lxd/network/openvswitch/ovn: Adds aclRuleDeleteAppendArgs, PortGroupPortSetACLRules and PortGroupPortClearACLRules functions
  • lxd/network/acl/acl/ovn: Renames ovnACLPriorityPortGroupDefaultReject to ovnACLPriorityPortGroupDefaultAction
  • lxd/network/acl/acl/ovn: Adds OVNApplyInstanceNICDefaultRules function
  • lxd/network/acl/acl/ovn: ovnApplyToPortGroup change default action to drop
  • lxd/network/driver/ovn: Error improvement
  • lxd/network/driver/ovn: Updates InstanceDevicePortSetup to setup NIC port default ACL rules
  • lxd/network/driver/ovn: Updates InstanceDevicePortDelete to use LogicalSwitchPortCleanup
  • lxd/network/driver/ovn: Adds security.acls.default.{in,e}gress.{action,logged} config key validation
  • lxd/network/driver/ovn: Adds default rule config change detection to Update
  • lxd/migrate: Set TrackProgress to true to track in-cluster progress information
  • lxd: Support for in-cluster progress information
  • lxc/move: Support for in-cluster progress information
  • lxd/device/nic/bridged: Check br_netfilter for security.ipv6_filtering loaded before clearing existing rules
  • lxd/device/nic/ovn: Typo in comment
  • lxd/network/openvswitch/ovn: Removes LogicalSwitchDHCPOptionsGetID
  • lxd/network/driver/ovn: Use client.LogicalSwitchDHCPOptionsGet in InstanceDevicePortSetup
  • lxd/network/openvswitch/ovn: Simplify LogicalSwitchDHCPOptionsDelete and make more efficient
  • lxd/network/openvswitch/ovn: Use more efficient o.LogicalSwitchDHCPOptionsDelete in LogicalSwitchDelete
  • lxd/network/openvswitch/ovn: Makes logicalSwitchDNSRecordsDelete more efficient
  • lxd/images: Skip update if image cannot be found
  • lxc/move: Support for quietFlag for in-cluster instance move
  • lxd/db/cluster: Extend list of entities
  • lxd/db/cluster: Fix entity URI for images
  • lxd/network/network/utils: Updates GetNeighbourIPs to return slice of NeighbourIP struct (with State)
  • lxd/device/nic/bridge: Updates State IP neighbour scanning to return valid state neighbours first
  • lxd/instance/drivers/driver/lxc: Adds statusCode function and update State to use it
  • lxd/instance/drivers/driver/lxc: Adds isRunningStatusCode function and updates IsRunning to use it
  • lxd/instance/drivers/driver/lxc: Adds renderState function and updates RenderState to use it
  • lxd/instance/drivers/driver/lxc: Updates RenderFull to use d.renderState
  • lxd/instance/drivers/driver/lxc: Updates IsFrozen to use d.statusCode()
  • lxd/instance/drivers/driver/lxc: Updates Render to use d.statusCode
  • lxd/instance/drivers/driver/lxc: Updates RenderState to use statusCode
  • lxd/instance/drivers/driver/qemu: Updates RenderState to use d.isRunningStatusCode
  • lxd/instance/drivers/driver/qemu: Updates IsRunning to use d.isRunningStatusCode
  • lxd/instance/drivers/driver/qemu: Updates IsFrozen to use d.statusCode
  • lxd/instance/drivers/driver/qemu: Reduce calls to statusCode in Render
  • lxd/instance/drivers/driver/qemu: Updates RenderFull to use d.renderState
  • lxd/instance/drivers/driver/qemu: Adds renderState and updates RenderState to use it
  • lxd/instances: Removes check for invalid state BROKEN in instancesShutdown
  • lxc/action: Allow --console with stop action.
  • i18n: Update translation templates
  • shared/api: Label Operation structs
  • lxd/operations: Add to swagger
  • doc/rest-api: Refresh swagger YAML
  • Makefile: Set GO111MODULE=off
  • lxd/operations: Cover public endpoints
  • shared/api: Fix cluster labeling
  • lxd: Add project arguments to swagger
  • lxd/swagger: Add background operations
  • shared/api: Label Image structs
  • lxd/images: Add to swagger
  • doc/rest-api: Refresh swagger YAML
  • lxd/instance/qemu: Disable net vectors on ccw
  • lxc/list: Handle dots in device keys
  • lxd/device/nic/bridged: Comment typo
  • lxd/network/driver/bridge: Error quoting
  • lxc/network/acl: Allow output of lxc network acl show to be used as input to lxc network acl edit
  • lxd/device: Ensure type field in NetworkInterface is specified first for liblxc
  • lxc/list: Add more instance filters
  • i18n: Update translation templates
  • Revert “doc: Added a Table of Content in /doc/index.md”
  • lxd/device/gpu_mdev: Simplify start logic
  • lxd/device/gpu_mdev: Support mdev on top of SR-IOV
  • lxc/list: Remove name filter
  • lxc/list: Rename state to inst
  • lxc/list: Cleanup comments
  • lxc/list: Pass state to filters
  • lxc/list: ipv4/ipv6 filters apply to state
  • lxc/list: Optimize filtering
  • lxc/list_test: Rename state to inst
  • lxc/list_test: Fix function signature
  • lxc/list_test: Fix name filter
  • lxc/list_test: Add state testing
  • i18n: Update translation templates
  • lxc/config: Add system-wide remotes
  • lxc/remote: Add system-wide remotes
  • lxc/info: Sort the mdev profiles
  • lxc-to-lxd: Fix test (type always goes first)
  • lxd/storage/drivers/driver/ceph/volumes: Don’t ignore unmount/unmap failures in DeleteVolume
  • doc: Remotes documentation
  • i18n: Update translation templates
  • test/suites/clustering: Retry removing node from database in test_clustering_remove_raft_node
  • lxd/gpu/physical: Fix compute-only passthrough
  • doc/instances: Mark mdev field as required
  • lxd/db/profiles: Cleanup arg names and errors in GetProfiles
  • lxd/api/internal: Adds internalImportRootDevicePopulate function
  • lxd/api/internal/test: Add tests for internalImportRootDevicePopulate
  • lxd/api/internal: Updates internalImport to use internalImportRootDevicePopulate
  • lxd/storage/errors: Removes ErrNotImplemented
  • lxd/storage/load: Return drivers.ErrNotSupported in GetPoolByInstance when storage pool doesn’t support instance type
  • lxd/storage/backend/lxd: drivers.ErrNotImplemented usage
  • lxd/storage/backend/mock: drivers.ErrNotImplemented usage
  • lxd/instance/drivers/driver/lxc: Check pool supports instance type in lxcCreate
  • lxd/instance/drivers/driver/qemu: Check pool supports instance type in qemuCreate
  • lxd/instance/drivers/driver/lxc: Use errors.Cause
  • lxd/instance/drivers/driver/qemu: Use errors.Cause
  • lxd/instance/drivers: Clearer pool load failure message
  • lxd/backup: Clearer pool load failure message
  • lxd/instance: Add revert package usage
  • lxd/instance: Clearer error messages
  • lxd/migrate/instance: storageDrivers.ErrNotImplemented usage
  • test/suites/backup: Add tests for checking lxd import profile root disk support
  • lxd/storage/backend/lxd: Check custom volume type is supported by storage pool
  • cluster: Larger Timeout to find leader
  • lxd/firewall/firewall/interface: Add NetworkSetup and remove feature specific network setup functions
  • lxd/firewall/drivers/driver/consts: Add network firewall Opts
  • lxd/firewall/drivers/drivers/nftables: Implement NetworkSetup and unexport per-feature setup functions
  • lxd/firewall/drivers/drivers/xtables: Implement NetworkSetup and unexport per-feature setup functions
  • lxd/network/driver/bridge: Switch to n.state.Firewall.NetworkSetup usage
  • doc/instance: Fix escaping
  • lxd/device/gpu: Add NVIDIA MIG support
  • doc/instances: Add GPU MIG
  • api: gpu_mig extension
  • doc/projects: Sort config keys
  • lxd/project: Add skipUnset
  • lxd/project: Refactor instance counting
  • api: project_usage
  • shared/api: Add ProjectState
  • lxd/project: Add getAggregateLimits
  • lxd/project: Add GetCurrentAllocations
  • lxd/projects: Add state endpoint
  • client: Add GetProjectState
  • lxc/project: Add info sub-command
  • i18n: Update translation templates
  • doc/rest-api: Add project state
  • doc/rest-api: Refresh swagger YAML
  • tests: Test for lxc project info
  • client: Fixes GetContainerLogfiles and GetInstanceLogfiles
  • doc/metadata: Adds remotes section
  • lxd/firewall/drivers/drivers/xtables: errors.Wrapf usage
  • lxd/firewall/drivers/drivers/xtables: Removes unused args from generateFilterIptablesRules
  • lxd/firewall/drivers/drivers/xtables: Adds iptablesChainCreate and iptablesChainExists functions
  • lxd/firewall/drivers/drivers/xtables: Moves NIC level security filtering rules into own chain
  • shared/validate: Allow uppercase letters in PCI addresses
  • shared/validate: Update unit tests for PCI addresses
  • lxd/instance/drivers: Restrict virtiofs to Intel architectures
  • lxd/device/disk: Restrict virtiofs to x86_64
  • lxd/cluster: Don’t copy all images on startup
  • tests: Disable test_clustering_image_replication
  • lxd/task: Add Hourly convenience function
  • lxd/images: Use Hourly() in autoUpdateImagesTask
  • lxd/firewall/firewall/interface: Adds delete and ipVersions slice args to NetworkClear
  • lxd/network/driver/bridge: n.state.Firewall.NetworkClear usage in setup
  • lxd/network/driver/bridge: firewall setup debug logging
  • lxd/network/driver/bridge: n.state.Firewall.NetworkClear usage in Stop
  • lxd/firewall/drivers/drivers/nftables: NetworkClear updated with new arguments
  • lxd/firewall/drivers/drivers/xtables: Updates networkSetupNICFilteringChain to use network-specific chain
  • lxd/firewall/drivers/drivers/xtables: Updates generateFilterIptablesRules to use network-specific chain
  • lxd/firewall/drivers/drivers/xtables: No need to use LookPath in iptablesChainCreate
  • lxd/firewall/drivers/drivers/xtables: Adds iptablesChainDelete function
  • lxd/firewall/drivers/drivers/xtables: NetworkClear updated to add new arguments
  • doc: Add aliases to snapshots.schedule
  • api: snapshots_schedule_aliases
  • lxd/util: Rework random functions
  • lxd/instance: Move NextSnapshotName
  • lxd/snapshot: Add snapshot aliases
  • lxd: Use snapshot aliases
  • lxd/operations: Fix possible NPE
  • lxd/instance: Add startup snapshot
  • shared/validate: Add IsCron validator
  • tests: Add snapshot scheduling
  • i18n: Update translations from weblate
  • lxd/storage/drivers/driver/zfs/volumes: Only mount instance filesystem volumes in postHook for CreateVolumeFromBackup
  • lxd/operations: Don’t wait for tokens
  • lxd/images: Run autoSyncImagesTask hourly
  • lxd/db/images: Add GetImages
  • lxd/images: Retrieve all images on sync
  • lxd/images: Use CopyImage() in autoSyncImages
  • lxd/daemon: Fix comment about autoSyncImages
  • lxd: Sync images on cluster node removal
  • lxd: Sync images when cluster.images_minimal_replica changes
  • tests: Re-enable image replication tests
  • lxd: Sync images on cluster node join
  • lxd/images: Pick a random source node for replication
  • lxd/operations: Export OperationClass type
  • client: Fix copy of VM snapshots

Try it for yourself

This new LXD release is already available for you to try on our demo service.

Downloads

The release tarballs can be found on our download page.

Binary builds are also available for:

  • Linux: snap install lxd
  • MacOS: brew install lxc
  • Windows: choco install lxc
8 Likes

Submitted to openSUSE. :smiley_cat:

3 Likes

Thanks!

On our side, it’s now in the stable channel for all distributions using the snap.

I have disabled ipv6 at work due to having network issues. After update lxdbr0 doesn’t work (and thus lxd in general). When I start fresh (snap remove -> install again) and run lxd init after answering initial questions (even with ipv6 CIDR set to none <- What IPv6 address should be used? (CIDR subnet notation, “auto” or “none”) [default=auto]: none)
I get the error:

Error: Failed to create network "lxdbr0" in project "default": Failed to setup firewall: Failed to run: ip6tables -w -t filter -I FORWARD -i lxdbr0 -j REJECT -m comment --comment generated for LXD network lxdbr0: ip6tables v1.6.1: can't initialize ip6tables table `filter': Address family not supported by protocol
Perhaps ip6tables or your kernel needs to be upgraded.

You opened a github issue

So we’ll track it there. Thanks

1 Like

Many thanks for quick response! I will follow the github issue from here.

To anyone having difficulties, there is a quick workaround that @tomp helped me achieved. You can follow the whole troubleshooting in the git issue tacker.

You should specify ipv6.firewall: "false" in your network (not NIC device) config for lxdbr0.

e.g.

config:
  ipv4.address: 10.1.100.1/24
  ipv4.nat: "true"
  ipv6.address: none
  ipv6.firewall: "false"

Ok, I can confirm this workaround works for the moment. And now I understand the full syntax for this setting. To summarize, here is my full preseed file in which I concatenated network and profiles setting:

lxd_preseed.yml

config: {}
storage_pools:
- config:
    source: mypool/lxd
    volatile.initial_source: mypool/lxd
    zfs.pool_name: mypool/lxd
  description: ""
  name: default
  driver: zfs
networks:
- name: lxdbr0
  type: bridge
  config:
    ipv4.address: 10.1.100.1/24
    ipv4.nat: "true"
    ipv6.address: none
    ipv6.firewall: "false"
profiles:
- name: default
  description: Default LXD profile
  config: 
    raw.idmap: "both 1000 1000"
    boot.autostart: "false"
    boot.autostart.delay: 5
    user.user-data: |
      #cloud-config
      package_upgrade: true
      runcmd:
        - <some commands>
      packages:
        - <packages>
      users:
        - name: myusername
          lock_passwd: false
          sudo: ALL=(ALL) NOPASSWD:ALL
          uid: "1000"
          shell: /usr/bin/zsh
          groups: sudo
          ssh_authorized_keys:
          - <mysshkeys>
  devices:
    root:
      path: /
      pool: default
      type: disk
    eth0:
      name: eth0
      network: lxdbr0
      type: nic


Then one shall run lxd init --preseed < lxd_preseed.yml to skip interactive settings.
When should we expect to see this fix in the snap?
Thank you!

1 Like