Measured on a VM with two disks, a CD-ROM and a NIC: 70 libvirt calls per
page, now 43. Within the page's cached_reads block the VM's state, the
host's info and domain capabilities, and each disk path's volume are read
once; the running and the persistent disk lists share a path's volume.
The ISO list takes the active pools in one call instead of looking each up
by name; it still refreshes them, so an image copied in by hand shows up.
The opt-in DRBD status waits at most 10 s for the compute's ssh, then
leaves the status unknown instead of holding the page.
The ARP lookup passed 3 as the address source; libvirt's
VIR_DOMAIN_INTERFACE_ADDRESSES_SRC_ARP is 2 and libvirt refused 3 as an
unknown source. The error was suppressed, so a VM without a guest agent
never showed an address.
libvirt keeps a VM's UEFI variables (NVRAM) and emulated TPM state in
files of its host. A live migration carries both: on the test hosts
(libvirt 11.10, QEMU 10.1) the destination got the same NVRAM content
once the VM ran and the same TPM state, and the source TPM state was
removed. Moving a shut-off VM moves only its definition, so neither
went along: on its first start on the destination libvirt made a new
NVRAM from the template and a new TPM. Boot entries and enrolled keys
are lost, and disks sealed to the TPM no longer unlock.
A shut-off VM whose definition has an NVRAM file or an emulated TPM is
now refused before anything changes: "Start the VM and migrate it
live." A passthrough or external TPM keeps no state on the host and
still moves. The check sits in moveto, not in the define/undefine
workaround, as libvirt's own offline migration does not carry the files
either. The keep-NVRAM and keep-TPM undefine flags of that workaround
are gone, as no VM that needs them reaches it.
Checked on the test hosts: a shut-off UEFI VM with a TPM is refused and
stays on the source with nothing defined on the destination; the same
VM migrates live. The new tests fail on the previous code.
A VM that changes its memory faster than the network carries it never
finishes a live migration without auto converge: measured with a guest
rewriting about 1.2 GB/s over an 8.5 Gbit/s link. The migration then
ran until gunicorn ended the request after 600 s. libvirt cancelled it
when the request went away and the VM stayed on the source, but the
user got an error page and nothing was logged.
The app now cancels a migration that has not finished after 540 s
(MIGRATION_TIME_LIMIT, below the 600 s of gunicorn.conf.py; a test
keeps it below). The failure is reported and logged like any other:
the migration was cancelled, the VM stays on the source host, and a
VM like that needs auto converge. A timer that fires as the migration
ends finishes before the compute locks are released, and an error
the cancel did not cause keeps its own message.
Migration needs the VM's disks on storage both hosts reach, and nothing
checked that. A VM with a disk the destination did not have failed
during a live migration; a shut-off VM's definition moved and could not
start there.
moveto now refuses the migration before anything changes, naming each
disk (target and source) that is not a volume of the destination's
active storage pools. Checked: the disks, CD-ROM images and block
devices of the live and the persistent definition, pool volumes, and
backing files, as the definition or the destination's volume lists
them (a shut-off VM's definition does not). Pools are refreshed before
a disk counts as missing, so a file another host made on shared
storage is found. Network disks and network backing images are
reached the same way from every host and are not checked; disk types
that cannot be checked (dir, nvme, ...) are refused.
This checks that the destination knows a volume, not that it is the
same file: shared storage should be a storage pool on each host. A
shared file outside any pool, or a block device no pool covers, is
refused.
moveto read the VM state for each flag and dropped options that did
not fit it, so a request could run as something else than asked; the
log then named the mode from the form ("live" or "offline"), not the
one used. The API did not log at all and answered "instance migrate is
started" after the migration had finished.
- moveto reads the state once. A running VM migrates live, or without
live paused for the transfer (non-live); a paused VM stays paused; a
shut-off VM moves offline. Live for a shut-off VM, offline for a
running or paused one, and any other state are refused before
anything changes. unsafe, auto converge (live only) and compressed
apply to a running VM.
- migrate_instance logs the outcome for the VM page and the API: the
mode and options used, or the error. The API answers
"instance is migrated (<mode>)" and accepts requests that leave out
an optional option.
- Post copy is gone from the form, the view and the API: only the
flag was set, and without migrateStartPostCopy, which the app never
called, the migration never switched to post copy.
libvirt's offline migration fails on some builds with "operation
failed: domain is no longer running" (RHEL-156800: Rocky/RHEL 9.8
libvirt 11.10.0-12 has the change that broke it, not the fix, which is
upstream in 12.2.0). A shut-off VM could not be migrated at all there.
A shut-off VM is only its definition, as the disks are on storage both
hosts share, so moveto now moves that itself:
- refused first when the VM has snapshots or a saved state, or the
destination already has a VM with its UUID or name;
- defined on the destination from the source's migratable XML, then
undefined on the source; NVRAM and TPM state stay on the source;
- after a failure the VM stays defined on one host: the destination
definition is removed only while the source still has its own, and
a VM another client started on the source meanwhile gets its
definition and autostart back there.
The destination's qemu migration hook does not run for it. Running and
paused VMs still migrate through libvirt.
Migration is client-driven: libvirt sends the VM's memory to the
destination's own host name unless a migration URI is given, and the
app never gave one. Where the hosts cannot resolve each other's names
(or the names point at a management network) every migration failed
with "Unable to resolve address '<host name>' service '49152'".
A compute now has an optional migration address (an IP address or a
host name, no scheme or port), editable in the compute form and the
API. A VM migrating to that compute is sent to tcp://<address>
(bracketed for IPv6). Empty, the default for existing computes, keeps
the previous behavior. The compute's host name is not reused, because
for SSH computes it can be an ssh_config alias.
The migration log entry names the address when one is set.
Migration left the VM defined on the source host unless "Delete
original" was ticked (VIR_MIGRATE_UNDEFINE_SOURCE). With the shared
storage migration needs, that definition points at the disks the
destination VM is running on: the VM showed on both hosts and could be
started (or autostarted) a second time against the same disks; only a
configured lock manager stops that.
moveto now always passes VIR_MIGRATE_UNDEFINE_SOURCE with
VIR_MIGRATE_PERSIST_DEST. The option is gone from the form, the view,
the API (MigrateSerializer.xml_del) and the clone auto-migration. It
removes the definition, not the disk files.
The VM page's getters read the domain XML on their own: one load made
86 libvirt calls for a real VM, 29 of them for the same three XML
documents (live, live with secrets, persistent) and 6 for the host
capabilities. Each call is a round trip to the host.
wvmInstance.cached_reads() is a block in which _XMLDesc (per flags
value) and get_cap_xml are read from libvirt once; it ends with the
block, even on an error, does not nest, and _defineXML drops it.
Outside a block every read goes to libvirt as before. Only the VM page
view, which does not change the VM, uses it, around its rendering.
Views that change the VM keep reading fresh XML: most changes (devices,
QoS, memory) do not go through _defineXML, and several views read,
change and define more than once in one request.
Measured on the test host: 86 -> 55 calls (XMLDesc 29 -> 3,
getCapabilities 6 -> 1) for one VM, 82 -> 54 for another; the page took
0.23-0.25 s before, 0.14-0.15 s after. The HTML of both pages is
identical before and after (CSRF tokens aside). A test renders the page
on libvirt's test driver through the real getters: 25 XML reads before,
one per kind now.
Each open VM page asks for its statistics every 10 seconds. For a
running VM the stats view measured the CPU, then each disk, then each
NIC, sleeping one second for each. With two disks and a NIC a request
held a sync gunicorn worker for over 4 s, so a few dozen open VM pages
could keep every worker busy.
wvmInstance.usage() replaces cpu_usage, disk_usage and net_usage, the
stats view's only callers: it reads all counters, sleeps one second
once and reads them again. Values and units are unchanged (CPU percent
of the host, disk bytes, network bits per second); compared against the
previous methods on the same counters, the results are identical.
Measured on the test host for a running VM with two disks and a NIC:
the stats request took 4.1-4.3 s before, 1.1-1.3 s now. The new test
fails on the previous code.
A VM owner who may manage the console (is_change and is_vnc) could set
its listen address to all interfaces and remove its password: a VNC
console reachable from the network without a password. Checked on a
test host: the owner's form changed the VM's listen address to 0.0.0.0
and then removed the password.
- The listen address field is disabled for users who are not
superusers; a disabled field ignores posted values. They can still
set or change the password and the keymap.
- A console whose persistent definition does not bind it to loopback
(or a Unix socket) only loses its password through an administrator.
A binding the XML does not pin down counts as not local: listen type
network, or no address (the host's default in qemu.conf).
On the test host, the owner's form now leaves the listen address alone
and is refused when it removes the password of a console listening on
0.0.0.0; with the console on localhost the password can be removed.
The new tests fail on the previous code.
attach_disk (new or existing volume, CD-ROM), add_guest_agent and
remove_guest_agent acted only when the status was running (1) or shut
off (5). On a paused VM they changed neither definition and the request
reported success. They now use the same rule as the NIC edits: the
running definition while the VM is active, paused included, and the
persistent one always. remove_guest_agent reads each definition from its
own XML, so a channel present only in the persistent one is removed too.
Unit tests check paused, running and stopped VMs (paused fails on the
previous code). The live effects test adds attach_disk and
add_guest_agent, and now also checks that edits meant for the running VM
reach it, which showed that an ejected CD-ROM keeps an empty <source>
in the live XML; NIC unplug is left to the guest tests, as these VMs
have no OS to acknowledge it.
Comments, docstrings and live test names cited item IDs of a roadmap
that is not part of the repository. They now say what is meant, or
nothing where the sentence stands without them. Test assertions and
behavior are unchanged.
- change_network detached the NIC and attached one built from scratch,
dropping its QoS, link state and other settings. It now edits the NIC
of the persistent definition in place and defines once; a running VM
gets the change at its next start, as the UI already says.
- add_network, delete_network and set_link_state did nothing on a
paused VM, and set_link_state rewrote the persistent NIC from the live
XML. They now update the running definition when the VM is active and
the persistent one always, each from its own XML.
- Boot and console edits used xml.etree, which renamed the namespace
prefixes of other tools' metadata (<app:info> became <ns0:info>); they
use lxml now.
- The NIC table and the clone form are filled from the persistent
definition (Instance.config_networks, config_disks), so a pending NIC
change is shown and can be edited again.
A live test runs 18 domain edits on stopped, running and paused VMs
whose definition has elements WebVirtCloud does not manage, and checks
for each the effect, that nothing else in the persistent definition
changed, and that persistent-only edits leave the running VM alone.
Clone:
- only shut-off VMs are cloned, as the UI already required: a running
guest keeps writing while its disks are copied; the definition comes
from the persistent XML;
- MAC addresses and disk destinations are checked before anything is
allocated;
- every volume the clone creates (NVRAM and disks) is tracked; if a
later copy or the definition fails, only those are deleted, and any
that cannot be deleted are named in the error. Before, a failed clone
left its earlier copies behind.
libvirtError(msg) reports the thread's last libvirt error instead of
msg when there is one, so WebVirtCloud's own errors could show an
earlier, unrelated failure. util.OperationError, a libvirtError
subclass that keeps its message, replaces those raises in vrtManager,
instances and nwfilters; 'except libvirtError' still catches it.
SPICE was removed earlier; its leftovers and the console type choice go
too:
- the console settings form, update_console and the backend no longer
offer or change the type; an invalid console form now shows its
errors instead of being ignored;
- the create form, the API serializer and create_instance no longer take
a graphics type, new VMs get VNC; get_graphics_types and the graphics
entries of the domain capabilities are gone;
- migration appsettings.0013 deletes QEMU_CONSOLE_DEFAULT_TYPE;
- a VM defined elsewhere with another graphics type gets "Console type
... is not supported".
The locale catalogs still reference the deleted SPICE templates in
source comments; regenerating them is left to the translation work.
The old code edited the live XML, called attachDevice and redefined the
domain from it: a stopped VM lost its VNC password, a paused VM crashed
with UnboundLocalError, an unknown CD-ROM was silently ignored (on a
running VM the last disk of the loop could be updated instead), and a
missing ISO raised AttributeError.
Media changes now use updateDeviceFlags on the requested CD-ROM, built
from the live and the persistent definition separately. A CD-ROM
missing from one definition, an unknown ISO, an ISO name present in
more than one pool, or a change that reached only the running VM is
reported to the user. The mount and unmount log messages were swapped.
Edits that redefine a domain read the wrong XML:
- boot menu/order, vCPU hotplug and QoS read XMLDesc(0), which has no
secrets, so every such edit, even on a stopped VM, removed the VNC
password from the definition;
- console, video, CPU, memory and title/description edits read the live
XML, so on a running VM pending changes (e.g. a new memory size) were
reverted and runtime state was written into the definition.
All of them now read VIR_DOMAIN_XML_INACTIVE | SECURE (PERSISTENT_XML).
The console setters take the first graphics device of that definition
instead of looking it up by the live type.
Disk resize no longer redefines the domain. A running VM's image files
grow through blockResize, since qemu-img cannot take the lock QEMU
holds; block volumes (e.g. LVM) are grown by the storage pool first and
QEMU is then told the new size.
Live tests reproduce the losses on a RHEL qemu-kvm host; unit tests
check the XML flags, the console setters and the resize paths in CI.
add_network built the interface XML with f-strings, so a posted model
or nwfilter could add arbitrary devices to the domain. It is built with
lxml now; a missing MAC is left to libvirt instead of address='None'.
Taking or reverting an internal snapshot redefined a pflash VM with a
rom loader around the operation. libvirt 10.10 rejects that definition,
so the request ended in a 500; where it was accepted, the snapshot
recorded the fake loader, and a failure between rewrite and restore
left the VM unable to boot.
The definition is no longer touched: libvirt decides, and its error is
shown (for a running VM with raw NVRAM: 'internal snapshots of a VM
with pflash based firmware require QCOW2 nvram format'). A stopped VM's
disk-only internal snapshot is taken, as libvirt allows it. Revert
errors are shown too. The external snapshot path keeps its rewrite for
now.
Live tests cover running and stopped UEFI VMs; both failed with a 500
on the previous code.
Instance.proxy looked the domain up by its stored name. After a rename
or a name swap outside WebVirtCloud, an operation on one VM changed
another; after a VM was replaced under the same name, operations hit
the new VM.
wvmInstance takes the UUID and uses lookupByUUIDString, with no
fallback to the name: a missing UUID fails instead of picking another
domain. Every caller that has an Instance passes it: proxy, console,
quota, post-migration autostart and the VDI URL. Migration uses the
domain the proxy already resolved. getvvfile and the gstfsd requests
use the host's current name of that UUID.
Live tests cover a name swap and a VM replaced under the same name;
both fail on the previous code.
Destroy:
- Stops every active state, not only running; a paused VM used to keep
running as a transient domain after the undefine.
- Undefines with MANAGED_SAVE and SNAPSHOTS_METADATA, before any disk is
deleted. A managed-save image used to make the undefine fail after the
disks were already gone; now a failed undefine deletes nothing.
- Keeps disks that another domain references, directly or as a backing
file anywhere in its chain (running domains' backingStore, and the
backingStore of the volumes in the active pools for stopped ones), and
says so; they used to be deleted with the VM.
Volume delete:
- Detaches from the live domain in every active state, including paused.
- Deletes the volume only after neither the live nor the persistent
definition has the disk (bounded wait for the guest to release it),
and refuses a volume another VM references. Otherwise the volume is
kept and the user is told why.
Both hold the compute exclusively for the usage scan and the deletion,
so another VM cannot attach the disk in between; the destroy
confirmation page (GET) takes no lock.
Live tests reproduce each case on a RHEL qemu-kvm host and fail on the
previous code; unit tests cover the destroy ordering and the locks.
The edit form posts net-*-<n> fields for NIC n and carries only the
new MAC. The backend read index 0 and looked the NIC up by the new MAC,
so editing a MAC detached nothing and attached a second NIC, and edits
of any NIC but the first were applied with empty values.
The form now also posts the current MAC (net-old-mac-<n>). The view
reads the submitted NIC's fields, validates the new MAC and reports
errors. The backend replaces the NIC with the old MAC, refuses an
unknown one, and on a failed attach restores the old NIC and raises
instead of reporting success.
The view crashed with a TypeError without a Referer header; it now
returns to the instance page, and a Referer from another site is
ignored.
Disk edits:
- edit_disk built a new <disk> from the form and replaced the old one,
losing what the form does not show (e.g. encryption and virtio
driver options); a bus change detached and re-attached the disk.
The disk of the persistent definition is now edited in place and the
domain defined once; a bus change picks a free device name and drops
the old address and virtio-only options. Pending persistent changes of
a running VM are kept.
- Clone omits empty disk devices, refuses a disk without a supported
source and destination, and checks the name before allocating storage.
Locks:
- Every VM change held its compute's exclusive lock, so one slow
operation blocked all VMs on that compute. A VM change now holds its
own VM lock plus a shared compute lock; reconciliation, migration and
cloning hold the compute exclusively (two clones of different VMs
must not pass the name check together).
- Quota-checked changes of one user (CPU, memory and disk resize,
clone) run one at a time under a per-user lock, also across computes,
so concurrent requests cannot all pass the check against the old
allocation. It is the outermost lock; superusers, who have no quota,
skip it.
- A busy lock answers 409 to the API and a message in the web UI. API
power and destroy actions return the view's error instead of success.
A disk without driver/@type could not be edited: the empty format was
rejected, and edit_disk would have written type=''. An empty format is
now accepted and edit_disk leaves the type attribute out.
The edit_volume bus change detached the disk and attached a new one built
as a file disk, so block and rbd disks lost their source and auth, and a
failed attach left the disk detached. Add wvmInstance.change_disk_bus,
which reuses the disk element and only changes <target> (dropping the old
bus address), and re-attaches the original disk if attaching fails.
edit_disk also rebuilt every disk with <source file=...>; it now keeps the
existing <source> and <auth> of block and network disks.
Disk and media mutation views (add_new_vol, add_existing_vol, edit_volume,
delete_vol, detach_vol, add_cdrom, detach_cdrom, mount_iso, unmount_iso)
were reachable by any owner with is_change, although the UI only shows the
Disk tab to superusers. An ordinary user could delete any volume on the
compute or attach arbitrary host files to their VM.
- Require superuser for all disk/media views, matching the UI.
- delete_vol resolves the volume from the VM's own disk XML by dev and
ignores the posted storage/name.
- Validate bus/cache/io/discard/detect_zeroes against the hypervisor's
lists, and format/serial/volume name against strict patterns.
- add_existing_vol only accepts volumes that exist in the selected pool.
- edit_volume only accepts a dev attached to the VM.
- vrtManager: escape all values interpolated into disk XML and pass dev
as an XPath variable in detach_disk/edit_disk.
* Add rest framework for API: First Commit
* modify some shell scripts to make variable references safer; modify some python scripts to reduce the code complexity and cyclomatic complexity of functions.
* Add REST API for some webvirtcloud functions. Instance list/delete/create, compute list/delete/create, storages-network list/retrieve. Add swagger and redoc for API interface
* update requirements
Co-authored-by: herengui <herengui@uniontech.com>
