cv4pve-report: baixar versions anteriors

What's new
- Capacity Planning section. One row per guest, node and storage with the numbers a sizing starts from: what is allocated, what is used on average, what is used at the peak, and how fast each storage fills up. No metric server or Grafana needed: it is built on the RRD data the nodes already keep (#85).
- Guests: vCPUs, CPU average and peak (% and cores), memory size, average and peak, disks size and used, network average and peak.
- Nodes: the same usage columns, plus the vCPUs and the memory assigned to the running guests with their ratio to the node (above 1 the node is overcommitted).
- Storage: size, used, free, growth per day and days to full.
- Averages come from the RRD samples read with the Average consolidation, peaks from a second read with Maximum: the highest of the averaged samples is not the peak. One more API call per guest and per node, only when the section is on.
- Turned on by the new setting CapacityPlanning.Enabled: off by default, on with --full. Each table needs the RrdData.Enabled of its scope (guests, nodes, storage) and uses its time frame.
- It is a summary, so it comes first: after Summary and Issues in Excel and in the Contents table, under Network Diagram in the HTML sidebar. JSON: capacity-planning.json, and filters.capacityPlanning in metadata.json.
Changed
- Documentation: new Capacity Planning page; Disks, Partitions and Snapshots moved from the VMs & containers page to the Storage page, as in the sidebar of the HTML report.
What's Changed
- docs: home page title for search results by @franklupo in #83
- docs: theme 2.6.2 by @franklupo in #84
- feat: Capacity Planning section with average and peak usage by @franklupo in #86
- chore: release 2.9.0 by @franklupo in #87
Full Changelog: v2.8.1...v2.9.0

Changed
- No em dashes in the report output. HTML page titles read Nodes · cv4pve-report, VM and container pages 100 (name), the footer cv4pve-report v2.8.1 · Corsinvest Srl · date; a missing Health value is -; Issues messages read 500 Reason: error (GET /path); network diagram tooltips read VM: 100 (name).
- Corsinvest.ProxmoxVE.Api 9.2.4
What's Changed
- docs: Connection page with response files, shorter Getting started by @franklupo in #77
- docs: "Options in a file" instead of "response files" by @franklupo in #78
- docs: AI assistants page and cv4pve-report skill by @franklupo in #79
- docs: shorter AI assistants page, no sandbox section, theme 2.6.0 by @franklupo in #80
- docs: replace em and en dashes in Markdown and docs site by @franklupo in #81
- chore: release 2.8.1 by @franklupo in #82
Full Changelog: v2.8.0...v2.8.1

Fixes
- Hostnames with dashes are kept in HTML and JSON file names. A node called pve-host01 used to produce pvehost01.html and pvehost01.json — the dash was dropped. The file name now matches the hostname you actually use.
- Links between HTML pages now work. Clicking a node, VM or container in an overview table opened a 404 instead of the matching detail page. The links go to the right page again.
- Two nodes with similar names no longer share the same report page. When two nodes ended up with the same file name (e.g. cc.01 and cc-01), the second one used to overwrite the first one's HTML/JSON file silently. Each section now gets its own file.
What's Changed
- chore: add unit tests + fix slug for hostnames with dashes by @franklupo in #56
- fix: HTML cross-section links + dedup HTML/JSON file paths by @franklupo in #57
- chore: bump version to 2.5.1 by @franklupo in #58
Full Changelog: v2.5.0...v2.5.1

What's new
- Health Score column on Nodes / VMs / Containers / Storages. Each row now carries a single 0–100 health number that summarises the resource's resource pressure (Node: weighted CPU/RAM/Disk; running VM/CT: weighted CPU/RAM; Storage: Disk usage). Higher is healthier. HTML renders it as a colour-coded badge (green ≥ 80, yellow ≥ 60, red below); Excel applies a green-yellow-red colour scale so the worst offenders pop visually and you can sort the column to surface them; JSON exposes the raw number under a health key. Stopped VMs / unavailable values are emitted as null / blank.
Fixes
- HTML detail pages now use a consistent Info heading. The first key/value block of every section (Cluster, per-node, …) is the page header; it now renders under the fixed Info heading like the VM/CT pages already did, instead of repeating the page title (Cluster, cc01). This mirrors the JSON writer's i == 0 → "info" rule, so the same block is named consistently across HTML and JSON.
- Excel Summary hyperlinks resolve to the exact sheet. The Contents table on the Summary sheet matched the target sheet by name prefix, so a section could link to a sibling that merely shares its prefix (e.g. Cluster could land on Cluster Access). It now prefers an exact match and falls back to a prefix search only for names truncated by Excel's 31-character sheet-name limit.
- HTML Health Score column now sorts numerically. The header was missing the data-type="number" hint, so clicking it sorted the score as text (1, 10, 100, 2…). It now sorts as a number.
What's Changed
- feat: Health Score column for Nodes / VMs / Containers / Storages by @franklupo in #52
- refactor: share Health Score via SDK + writer polish by @franklupo in #54
- chore: bump version to 2.5.0 by @franklupo in #55
Full Changelog: v2.4.0...v2.5.0

What's new
- New JSON output format — --format Json. A third rendering alongside Excel and HTML, designed for automation, scripting and integrations. Produces a single zip containing one file per section (cluster.json, nodes.json, vms.json, storages.json, …) plus per-resource detail under nodes/<name>.json, vms/<id>.json and containers/<id>.json, same logical layout as the Excel sheets and HTML pages. A metadata.json carries the schema version, timestamp, application info, the filter set and a per-section generation log. The network topology SVG is bundled in the same zip as network-diagram.svg. Keys are camelCase with stable shapes (overview files are flat arrays of rows; detail files are objects with a fixed info block plus one entry per table), so consumers can navigate the dataset without knowing resource names in advance. Full reference in docs/format-json.md with file layout, naming conventions, jq recipes and a snapshot-diff workflow.
- CD-ROM and cloud-init drives now listed in the Disks sheet. The per-VM disk inventory used to skip ide2-style CD-ROMs and cloudinit drives, only showing real block devices. They now appear alongside the rest with a new Kind column (Disk, CDRom, CloudInit) so you can filter on what's actually a backed-up disk vs. a transient mount. Thanks for the report (#39).
What's Changed
- feat: include CD-ROM and cloud-init drives in the Disks sheet by @franklupo in #39
- feat: JSON output format (--format Json) by @franklupo in #40
- chore: bump version to 2.3.0 by @franklupo in #41
Full Changelog: v2.2.1...v2.3.0

Fixes
- EFI disk and TPM state now listed in the Disks sheet. efidisk0 and tpmstate0 were missing from the per-VM disk inventory because the underlying SDK consumed them into dedicated typed properties instead of leaving them in the generic disk dictionary. They now appear alongside scsi0, virtio0, etc. with the same columns (storage, format, size). Thanks @shaundeeb for the report (#37).
- SDN vnets no longer make VMs disappear from the network diagram. When a VM was attached to an SDN vnet that wasn't materialised in the node's /etc/network/interfaces (typical of zones whose apply hasn't been propagated, Simple/EVPN zones, or datacenter-only vnets), the diagram had no bridge to draw the VM on and the VM was silently dropped. The diagram now learns about SDN vnets from the cluster-wide config and injects the missing bridges as synthetic boxes on every node where the zone is active. Thanks @janrenard for the report (#36).
What's new
- Network sheet now includes an SDN Vnets table. Between Nodes Networks and VM Networks there is now a third table listing every SDN vnet with its zone, zone type, parent bridge, VLAN tag, alias and the nodes where it's active. Same data was already available in the Cluster sheet but it lives natively here too now, so the network view is self-contained.
What's Changed
- docs: unify format-xlsx/format-html structure + document Issues page by @franklupo in #35
- fix: efi/tpm disks listed, SDN vnets drawn in diagram (closes #37, #36) by @franklupo in #38
Full Changelog: v2.2.0...v2.2.1

What's new
- Resilient collection — broken endpoints no longer abort the report. A single failing Proxmox API call (a storage with a corrupt RRD file, a node returning 500, missing permissions on a sub-resource) used to terminate the whole report. Failures are now collected into a dedicated Issues page that lists every problem with severity, section, the full Proxmox error message and the API endpoint that returned it. 501 Not Implemented (endpoint missing in older PVE versions) stays silent — everything else surfaces as a Warning you can act on. Thanks @janrenard for the report (#33).
- Issues page — second tab after Summary / second link after Home. The Issues page only appears when there is something to report: a dedicated sheet in Excel positioned right after Summary, and the second sidebar link after Home in HTML. Issue rows are hyperlinked back to the relevant detail page (the failing VM, node, storage, or the global section) so you can jump straight to the context where the failure happened.
- Cover/Summary now shows a Description column. Both formats list every section in the Contents table with a one-line description of what's inside — consistent between Excel and HTML. Issue counts and per-section durations are unchanged.
What's Changed
- chore: writers refactor (unified ColumnKind, ZipHelpers, BlockFormat) and CSS cleanup by @franklupo in #32
- feat: resilient collection + Issues page (fixes #33) by @franklupo in #34
Full Changelog: v2.1.0...v2.2.0

Breaking changes
- All formats now produce a single .zip. Excel output used to be a pair of loose files (Report_*.xlsx + Report_*.svg); it is now packed into one Report_*.zip containing report.xlsx and network-diagram.svg. Same logical content, more portable: easier to share, attach to email/tickets, archive, or upload as a CI artifact — and ready for the upcoming --only flag that will produce multiple workbooks per run. To use the workbook, extract the zip and open report.xlsx as before. The --output flag now always writes a .zip; the extension is appended automatically if missing.
- Settings renamed — Sheet suffix dropped from section toggles. With three rendering formats (Excel, HTML, JSON-in-progress), the Sheet suffix in settings names was misleading: these flags toggle a section of the report, not a spreadsheet sheet. Rename your existing settings.json if you maintain one:
- Cluster.IncludeSheet → Cluster.Include
- Cluster.IncludeTasksSheet → Cluster.IncludeTasks
- Node.IncludeReplicationSheet → Node.IncludeReplication
- Guest.IncludeSnapshotsSheet → Guest.IncludeSnapshots
- Guest.IncludeDisksSheet → Guest.IncludeDisks
- Guest.IncludePartitionsSheet → Guest.IncludePartitions
- Storage.IncludeContentSheet → Storage.IncludeContent
- Storage.IncludeBackupsSheet → Storage.IncludeBackups
What's new
- HTML — per-column filter. Headers backed by short text, flag or hyperlinked values (Node, VmId, Status, Type, …) now carry a hover-revealed funnel icon. Click it to open a dedicated filter input under the column header — the global filter and the per-column filters combine with AND. The big winner is high-cardinality pages like Snapshots (~250 rows for 23 VMs) and RRD Guests (~1600 rows): drilling down to a single VM no longer requires scanning through false positives.
- HTML — match-mode toggle (~ / =). Both the global filter and each per-column filter expose a small toggle on the left of the input. ~ is the existing case-insensitive contains match (default). = is exact whole-cell match — essential for short numeric ids like a VM id where 100 would otherwise also match 1014, 260509100047, IP suffixes, ports, etc. The toggle re-applies the filter immediately so you can flip mode mid-query.
- HTML — three-state sort. Clicking a sortable header now cycles original → ascending → descending → original. The third click restores the order the engine emitted (e.g. nodes by hostname, snapshots by date) instead of leaving the table sorted by whichever column you last touched.
Fixes
- Per-sheet Index restored on Excel detail sheets. The clickable index of tables at the top of every node/VM/container detail sheet (lost during the 2.0.0 writers refactor) is back. It now uses a 2-column compact layout — much friendlier on resources with many tables (VM detail with QEMU agent, node detail with full disk/SMART/APT data) — and sits right after the resource identity block instead of at the very top of the sheet.
- HTML — header alignment matches column type. Numeric column headers (VmId, Size GB, MaxCount, …) are now right-aligned like their cells; flag headers are centred. A specificity bug introduced with the writers refactor was forcing every header to the left, breaking the visual axis on data-heavy pages.
- HTML — global filter input on the cover page. The "Contents" table on index.html lacked the .table-scroll wrapper used elsewhere, which collapsed the gap between the row count and the filter input above it. Wrapped consistently with the rest of the report.
Documentation
- Settings reference moved to a dedicated guide. The annotated settings.json example, the skip-heavy-sections cheatsheet and the performance tuning notes (MaxParallelRequests, ApiTimeout, QemuAgentTimeout) are now in docs/settings.md. The README keeps the profile comparison matrix inline (it's the decisional content readers actually want at a glance) and links to the full reference.
- Per-format reference guides. docs/format-xlsx.md and docs/format-html.md now host the lon

Fixes
- Reports no longer crash on VM/CT detail pages — a leftover code path in v2.0.0 still failed when a VM or container had a minimal config without extra fields. The detail page now renders cleanly in those cases. Thanks again @janrenard for spotting it (#24).
What's Changed
- fix: 2.0.1 — VM/CT detail crash on minimal configs by @franklupo in #30
Full Changelog: v2.0.0...v2.0.1

What's new
- HTML output — pick --format Html at the command line to get the report as a self-contained zipped website. Extract the zip and open index.html in any browser — works fully offline, no server needed. Built for large clusters (tested with 2700+ VMs), with a sidebar to navigate every node/VM/container, click-to-sort and per-table search, light/dark theme, and an Export button on every detail page that lets you share a single page as a stand-alone file (ready to email, paste into a ticket or drop on a wiki). The Excel output (--format Xlsx, the default) is unchanged.
Fixes
- Reports no longer crash on VMs with empty cloud-init values — a VM whose config has an empty cicustom: field used to abort the whole report. The failing VM is now skipped cleanly. Thanks @janrenard for the report (#24).
- Reports no longer crash on user accounts with very long expiration dates — accounts set to expire after the year 2038 caused the report to fail. All time-based fields (account expiry, scheduled next run, replication times, RRD timestamps, cluster log time…) now handle dates well into the future. Thanks @LordXearo for the report (#25).
- Reports no longer crash when the firewall log is empty — an empty firewall log used to abort the report. Empty logs (firewall, system journal, task log, replication log) now produce an empty section instead. Thanks @LordXearo for the report (#26).
- Clearer error messages — when a VM or container config fails to load, the error now identifies the offending guest (id, node and name) instead of a bare crash.
Documentation
- README reorganised with a dedicated Output Formats section: Excel and HTML side-by-side, so you can pick the one that fits your workflow at a glance.
What's Changed
- feat: writers abstraction + HTML report format by @franklupo in #27
- release: 2.0.0 — HTML output, large-cluster support and fixes by @franklupo in #29
Full Changelog: v1.8.1...v2.0.0

Fixes
- Compatibility with older PVE clusters — the /cluster/mapping/dir endpoint was added in newer Proxmox VE releases; older clusters (e.g. 8.3.x) returned 501 Not Implemented and crashed the report. The fetch now degrades gracefully: the Mapping Dir sheet is simply empty when the endpoint isn't available, and the rest of the report is unaffected. Thanks @janrenard for the report (#20).
What's Changed
- fix: handle 501 Not Implemented for /cluster/mapping/dir on older PVE by @franklupo in #22
- chore: release 1.8.1 by @franklupo in #23
Full Changelog: v1.8.0...v1.8.1

What's new
- Network topology diagram — every export now produces an SVG diagram next to the Excel file, with the same name. For each Proxmox node it shows the full network path from physical NICs to bonds, bridges, firewall/router VMs, internal bridges and the VMs/CTs they serve, plus a dedicated row for network storage (NFS, CIFS, PBS, iSCSI, Ceph, RBD, GlusterFS). Open it in any browser — colours and arrows make routing, multi-homed gateways and inactive interfaces immediately visible.
Fixes
- VM data more resilient — when a single QEMU agent call fails (e.g. get-fsinfo on OPNsense), the rest of the agent data (hostname, OS info, network) is still collected instead of being lost.
- Clearer error messages — when the QEMU agent fails the Hostname column now shows the actual reason, not just "Agent not running".
Documentation
- New page docs/network-diagram.md with the legend, layout and how to read the diagram.
- A sample diagram is included in the repo at docs/network-diagram.svg (cluster anonymised with example data).
- README updated with a preview of the new diagram.
What's Changed
- feat: network topology SVG diagram for v1.8.0 by @franklupo in #19
Full Changelog: v1.7.0...v1.8.0

What's new
- Host file in node detail — the host name resolution table (/etc/hosts) is now included in each node's detail sheet, right after the network interfaces
- Offline nodes — nodes that are offline are now skipped gracefully instead of producing errors in the report
What's Changed
- feat: /etc/hosts in node detail, skip offline nodes, RunParallelAsync for v1.7.0 by @franklupo in #18
Full Changelog: v1.6.0...v1.7.0

Fixes
- Disks sheet — storage type and usage now show correctly for VMs with disks spread across multiple storages
- Nodes Networks sheet — the Node column was always empty; it now shows the correct node for each network interface
- SSL certificate expiry — days until expiry no longer crashes for certificates with no expiry date
- Replication sheet — the Summary sheet now shows the correct number of replication jobs instead of the number of nodes
- Cluster sheet — fixed a duplicate API call when loading the Two-Factor Authentication table
Performance
- Firewall — rules, aliases and IP sets are now fetched in parallel for both cluster and each VM/CT, reducing report generation time on large clusters
- Replication — replication status is now fetched from all nodes in parallel instead of one at a time
- Storage Content — content and backup lists are now fetched and written in a single parallel pass
What's Changed
- fix: bug fixes, performance improvements and code cleanup (#17) by @franklupo in #17
Full Changelog: v1.5.0...v1.6.0

What's new
- Response files — host, username and password can be saved in a config.rsp file and passed as @config.rsp, so you don't have to repeat credentials on every run
- Generation stats in Summary — the Summary sheet now shows how many rows each section produced and how long it took, plus a total duration at the bottom
- ApiTimeout setting — new option to set the HTTP timeout in seconds for slow or high-latency clusters; leave at 0 to use the default (100 s)
- QemuAgentTimeout setting — new option to control how long to wait for the QEMU guest agent before giving up; default is 3 seconds
- Firewall log now opt-in per section — Node.Detail.IncludeFirewallLog and Guest.Detail.IncludeFirewallLog let you enable firewall logs independently for nodes and VMs/CTs
Fixes
- Firewall log no longer crashes the report when the API returns no data
- Sys.Modify permission added to the required permissions documentation (needed for APT repositories, updates and package versions)
Performance
- Nodes fetched in parallel — all nodes are now queried at the same time (status, version, subscription, DNS, time, network), the same way VMs and containers already worked
- Node detail queries parallelized — for each node: services and SSL certificates, APT repositories/updates/versions, and directory/ZFS pools are now fetched simultaneously instead of one after another
- Network data written once — node network interface rows are collected during node processing and written to the Network sheet in a single pass, instead of incrementally
What's Changed
- refactor: consolidate packaging metadata into single config file by @franklupo in #13
- fix: handle null firewall logs and add Sys.Modify to required permissions by @franklupo in #14
- feat: add IncludeFirewallLog flag to node and guest detail settings by @franklupo in #15
- feat: parallelise all API calls and add performance tuning docs by @franklupo in #16
Full Changelog: v1.4.0...v1.5.0

What's new
- Custom settings — generate a settings.json with create-settings, edit it, and pass it with --settings-file to fully control which sheets are generated
- Cluster sheet can be disabled — new Cluster.IncludeSheet flag to skip the Cluster sheet entirely
- Per-node detail sheets can be disabled — Node.Detail.Enabled = false skips all per-node detail sheets, useful on large clusters
- Per-VM/CT detail sheets can be disabled — Guest.Detail.Enabled = false skips all per-VM/CT detail sheets, the most impactful option for clusters with hundreds of VMs
- Individual global sheets can be toggled — Guest.IncludeDisksSheet, Guest.IncludeSnapshotsSheet, Guest.IncludePartitionsSheet to skip specific sheets
Changes
- Settings restructured: node detail options moved under Node.Detail (disk, tasks, APT)
- Settings restructured: guest detail options moved under Guest.Detail (tasks)
- MaxParallelRequests unified to a single global setting (previously split per-section)
- IncludeReplication renamed to IncludeReplicationSheet, IncludeTasks renamed to IncludeTasksSheet for consistency
- Firewall settings simplified: LogMaxCount → MaxCount, LogSince → Since, LogUntil → Until
- Fast profile now disables detail sheets and firewall — significantly faster on large clusters
- README reorganized: profiles and customization are now front and center
What's Changed
- feat: release 1.4.0 by @franklupo in #12
Full Changelog: v1.3.0...v1.4.0

What's new
- Network sheet — new dedicated sheet with a complete network inventory across the entire cluster: all node interfaces and all VM/CT network cards with MAC address, bridge, VLAN, IP addresses and OS info in one place
- Disks sheet — new dedicated sheet with VM/CT disk inventory: storage, size, cache, backup flag, unused flag, mount point, passthrough
- VM IP addresses — IP addresses from the QEMU agent are now visible directly in the VM overview and network tables
- VM overview — new columns: Networks (MAC + bridge/VLAN), IP Addresses, Hostname
- VM network detail — additional fields: Trunks, Disconnect, LinkDown, Gateway, Gateway6
- VM disk detail — additional fields: MountPoint, MountSourcePath, Passthrough
- Node network detail — full IPv6 fields now included: Cidr6, Address6, Netmask6, Gateway6, Method6
- Node disks — reorganized settings: SettingsDisk class with Enabled, IncludeSmartData, IncludeZfs, IncludeDirectory
- Node detail — new optional tables: Directory mount points, ZFS pools with vdev tree
- APT Repositories — new IncludeAptRepositories option to show configured APT repositories in node detail
- Cluster Audit Log — cluster event log can now be included in the Cluster sheet with OnlyErrors and MaxCount filters
- SkipEmptyCollections — new global setting to skip empty collections (e.g., no snapshots) from the report
- README — added "Where cv4pve-report fits" section with RVTools comparison table
Changes
- Version bumped to 1.2.0
- Settings reorganization: disk-related settings moved from SettingsNode to dedicated SettingsDisk class
- Audit Log settings enhanced with OnlyErrors filter for severity-based filtering
Fixes
- HA Groups correctly skipped on PVE 9 and later where the API endpoint was removed
What's Changed
- fix: skip HA Groups on PVE 9+ (closes #2) by @franklupo in #3
- feat: release 1.2.0 by @franklupo in #5
Full Changelog: v1.1.0...v1.2.0

What's new
- HA (High Availability) — the Cluster sheet now includes HA resources, groups and current status
- Resource Pools — the Cluster sheet now lists all pools with their members (VMs, CTs and storages)
- SDN Ipams and Subnets — SDN section now also covers IP address management and subnets per vnet
- Syslog — each node detail sheet can now include the system log (disabled by default, enabled in Full profile)
- Firewall logs — node and VM/CT firewall logs now support date range and line count filters
- Task filters — tasks on nodes and VMs/CTs can now be filtered by errors only, limited by count, and filtered by source (node only, all, active)
- Full profile — now includes syslog and firewall logs limited to the last 7 days (1000 lines max)
What's Changed
- feat: release 1.1.0 — HA, Pools, SDN, Syslog, task/firewall filters by @franklupo in #1
New Contributors
- @franklupo made their first contribution in #1
Full Changelog: v1.0.0...v1.1.0