Perch draws your estate as one interactive graph — cloud → host → VM → network, with operational overlays for compliance, backup, monitoring and recent activity on top. It rides entirely on data Morpheus has already synced: no agents of its own, no inventory of its own, and no calls outside your appliance unless you switch them on.
It is read-only in the strict sense. Perch never writes to Morpheus, never provisions, and never changes a host. Uninstalling it removes everything it was.
Every layer is labelled with where its data comes from, because that is what determines its cost. Everything that leaves the Morpheus database is off by default.
| Layer | Source | Default | Surfaces |
|---|---|---|---|
| Core — clouds, hosts, VMs, networks | Local SQL | on | all |
| Routers | Local SQL | on | all |
| Security groups + rules | Local SQL | on | all |
| Kubernetes pods + services | Local SQL | on | all |
| Compliance overlay | Local SQL | on | all |
| Backup overlay | Local SQL | on | all |
| Monitoring overlay | Local SQL | on | all |
| Recent activity overlay | Local SQL | on | all |
| Datastores + volumes | Local SQL | off | all |
| Hardware devices (GPU passthrough) | Local SQL | off | all |
| Host traffic volume | SSH | off | report, instance, server, cluster |
| Active host flows | SSH | off | report, instance, server, cluster |
| Tracked connections | SSH | off | report, instance, server, cluster |
| Storage paths (multipath / iSCSI) | SSH | off | report, instance, server, cluster |
| LLDP neighbours | SSH + lldpd |
off | report, instance, server, cluster |
| Contrail flows / Apstra fabric / flow replay | External REST | off | all |
The four surfaces beyond the report are Topology tabs on instance, server, cluster and network detail pages, each scoped to what that object touches.
The database layers should work on any Morpheus 9.0.1+. They read tables Morpheus itself populates.
The host-telemetry layers need libvirt and Open vSwitch. On VMware, Nutanix, Hyper-V or Linux-bridge KVM they cannot work as built — and Perch says so explicitly rather than rendering an empty layer.
Verified against exactly one environment: one Morpheus 9.0 appliance, one private cloud, four KVM/MVM hypervisors on Open vSwitch, roughly a dozen servers and a handful of networks.
That is the honest scope. Every threshold, every default, every type-code mapping that picks an icon was chosen by looking at that estate. When something looks wrong on yours, the first question is usually "is this a fault, or is my estate simply not the one this was tuned against?" — see LIMITATIONS.md and docs/SCHEMA.md.
A second cloud type — a standalone VMware ESXi host — has been handled as designed but not yet verified against a live ESXi cloud.
sha256sum -c perch-morpheus-plugin-<version>.jar.sha256Administration → Integrations → Plugins → Upload, select the jar, done. No appliance restart, no service to bounce.
Full instructions, plugin settings, sudoers and lldpd setup:
INSTALL.md.
Perch does not filter by tenant. It reads the Morpheus database directly, through a connection that performs no tenant evaluation, so it does not inherit Morpheus RBAC. Any user who can open a Perch surface sees the whole estate that surface returns.
getMasterOnly() is the only control. If you run sub-tenants, read
SECURITY.md first.
Perch has been verified on one estate, so a report that it worked on yours is as useful as a bug — it is the only way the "verified on" list above ever grows.
Five questions, and one-line answers are plenty:
- Hypervisor / cloud type — VMware, Nutanix, Hyper-V, KVM, AWS, Azure, mixed?
- Scale — how many servers, networks, clouds?
- Multi-tenant? Perch does not filter by tenant; if you run sub-tenants, say so.
- Which layers were empty, and did you expect data there?
- Anything that looked wrong but did not error — a wrong count, an odd glyph, a node in the wrong table. Silent wrongness is the most valuable thing to receive and the hardest to find from logs.
→ Open an "it worked" issue, or see FEEDBACK.md for what to send and how to send it without publishing your infrastructure.
| INSTALL.md | Install, settings, privileges, lldpd, SCAP |
| LIMITATIONS.md | What it cannot do, and why |
| SECURITY.md | Tenancy, what is never read, reporting a vulnerability |
| docs/ARCHITECTURE.md | How it works and what it costs |
| docs/SCHEMA.md | What it reads — for filing schema-difference issues |
| docs/CONFIGURATION.md | Every setting and run option |
| docs/DIAGNOSTICS.md | What to collect when something is wrong |
| docs/TROUBLESHOOTING.md | Symptom → cause → which diagnostic |
| CHANGELOG.md | What changed, per release |
Apache 2.0 — see LICENSE and NOTICE. Perch is a community plugin. It is not an HPE or Morpheus product and carries no support from either.