VmService.GetVms(node: null) now issues a single GET cluster/resources?type=vm
instead of GET /nodes followed by one GET /nodes/{n}/qemu per node -- 17 round
trips down to 1 on a 16-node cluster. VmService.GetVm(session, node, vmid) now
fetches nodes/{node}/qemu/{vmid}/status/current directly instead of listing
the whole node and filtering client-side. GetPveVmCmdlet's -Detailed path
shares one IPveHttpClient across the enrichment loop and calls WriteObject
per VM as it is enriched, instead of materializing the whole list first and
opening a fresh client per VM.
cluster/resources's "type=vm" filter returns both qemu and lxc rows (per the
PVE OpenAPI spec), so the mapping filters to type=="qemu" explicitly. The
resources endpoint does not carry QmpStatus/Pid/AgentStatus, so those stay
null until -Detailed or GetVm enrich from status/current; the default table
view (VmId/Name/EffectiveStatus/Node/CpuCount/MaxMem/Uptime) is unaffected.
GetVm's status/current call converts a 404 or 500 into the pre-existing
InvalidOperationException("not found") contract that ImportPveOvaCmdlet
depends on, but 502/503/504 propagate unchanged since PveHttpClient wraps a
connectivity failure as 503 -- a down node must not read as a missing VM.
Removing the old per-node fan-out (issue #142's node-skip behavior) means
onNodeSkipped is no longer invoked for the all-nodes path: a single
cluster/resources call has no per-node failure to report, so a
node-unreachable condition now surfaces via that node's rows rather than a
WriteWarning. The parameter stays on VmService.GetVms/TemplateService.GetTemplates
for source compatibility with the cmdlets that still wire it.
TemplateService.GetTemplates and its tests are also updated since it
delegates to VmService.GetVms and inherited the same per-node fan-out.
Co-authored-by: goodolclint-claude[bot] <323206664+goodolclint-claude[bot]@users.noreply.github.com>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* refactor: one task-response parser, and the dead code #154 names (part A)
Unifies the 8 byte-similar private ParseTask methods in BackupService,
ContainerService, NetworkService, NodeService, SnapshotService,
StorageService, TemplateService, and VmService into one shared
PveTaskResponse.Parse(json, node) utility, on the variant that stamps
Status = "running" for a bare-UPID response. BackupService, NodeService,
TemplateService, and VmService previously left Status null for that case;
their non--Wait task output now reports "running" like the other four
services already did.
Removes the duplicate ClusterConfigService.GetClusterStatus in favor of
ClusterService's; ClusterConfigService now holds a ClusterService built
from the same injected/default client, and WaitForQuorum and
Get-PveClusterStatus go through it.
Removes dead code named in issue #154 and its fold-in comments: the
never-called PveHttpClient/IPveHttpClient sync wrappers Put/Delete, the
never-thrown PveAuthenticationException, a #pragma around an
already-nullable field, the unreferenced TestHelper mock-handler helpers,
three hand-rolled version-warning blocks (now PveCmdletBase.WarnIfBelowVersion),
an unreachable catch(HttpRequestException) arm in WaitForStatusTransition,
dead ExitStatus-checking branches in ImportPveOvaCmdlet after WaitForTask
(which already throws on failure), and an unreachable int branch in
ApiValueHelper.IsExited.
Fixes the WaitForStatusTransition catch removal's premise: PveHttpClient
read the response body outside its HttpRequestException try block, so a
mid-body stream drop could still escape unwrapped. Moves the body read
inside the try so every HttpRequestException the client can throw becomes
a PveApiException, matching what the removed catch assumed.
Part of #154.
* Add service files: BackupService, ClusterConfigService, ContainerService, NetworkService
* Add service files: NodeService, SnapshotService, StorageService, TemplateService
* Add VmService and Utilities files
* Remove unused PveAuthenticationException (never thrown, caught, or tested)
* Add cmdlet files: GetPveClusterStatus, SDN subnets, PveCmdletBase, SendPveFile, ImportPveOva
* Add test files: BackupServiceTests, ClusterConfigServiceTests, NodeServiceTests
* Add remaining test files: TemplateServiceTests, VmServiceTests, TestHelper, ApiValueHelperTests, PveTaskResponseTests
* Add VmServiceTests
---------
Co-authored-by: goodolclint-claude[bot] <323206664+goodolclint-claude[bot]@users.noreply.github.com>
Get-PveCloudInitConfig and Get-PveTemplate each built their own PveHttpClient
inline. They now go through CloudInitService and TemplateService, which is
offline-testable per ADR 0021.
Get-PveCloudInitConfig's shipped output is the full PveVmConfig (per its own
synopsis), not the CI-key-filtered PveCloudInitConfig that
CloudInitService.GetCloudInitConfig already returns for a different purpose.
CloudInitService gains GetFullVmConfig, a thin delegation to the existing
VmService.GetVmConfig (already used by Get-PveVmConfig) rather than a
parallel HTTP implementation.
Get-PveTemplate's TemplateService.GetTemplates now forwards VmService.GetVms's
onNodeSkipped callback, matching Get-PveVm/Get-PveContainer, so an unreachable
node during the all-nodes listing produces a warning instead of a silently
short result. Two review-found regressions were fixed before commit: an empty
-Node value used to mean "query all nodes" and now does again (it had
started reaching the API literally as nodes//qemu), and the per-node warning
above.
Co-authored-by: goodolclint-claude[bot] <323206664+goodolclint-claude[bot]@users.noreply.github.com>
Every service method built and disposed its own PveHttpClient, and each
client owned a fresh HttpClientHandler, so every API call was a TCP connect
plus a TLS handshake and WaitForTask paid that 300 times over a ten-minute
wait.
PveServiceBase now owns the injected-or-fresh client lifetime behind one
Invoke helper; the 201 hand-written try/finally blocks across the 16
services collapse to calls on it, and the nested NodeService/VmService
instances receive the injected client. PveHttpClient takes its handler
from a process-wide PveHandlerCache keyed on (host, port,
skipCertificateCheck) and never disposes it, so the connection pool
outlives any one client. WaitForTask holds one client for the whole wait
and, when no pollInterval is supplied, backs off from 1 s toward a 10 s
cap, never sleeping past the deadline.
Co-authored-by: goodolclint-claude[bot] <323206664+goodolclint-claude[bot]@users.noreply.github.com>
- CS8604: Use cmdlet's own Node property instead of nullable task.Node
in WaitForTask calls across all VM cmdlets
- CS8618: Make PveSession.ServerVersion nullable (PveVersion?) since it
is set post-construction by PveAuthenticator
- CS8603: Add null-forgiving operator where IsNullOrWhiteSpace guards
guarantee non-null in PveHttpClient
- CS1573: Add XML param documentation to all public service methods
- xUnit1012: Make test parameters nullable (string?) for null test cases
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add service classes for nodes, VMs, containers, storage, snapshots,
network/SDN, users/roles/permissions, templates, cloud-init, tasks,
and cluster. SDN services enforce PVE 8.0+ version guard. Task service
supports polling with configurable timeout and progress callbacks.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
goodolclint-claude[bot]