Storage and NFS¶

Each zPod's zcore runs an NFS filer. The deploy gives it a system disk and a filer disk, which is mounted as /FILER/STORAGE01 and holds the shares the nested hosts use (NFS-01, NFS-VCD, VCF-BACKUPS). Storage self-service lets you add disks to the zcore VM, turn them into more storages, organize folders on them, and export those folders over NFS to the clients you choose.

There are three layers, each with its own zcli command group:

You can also do all of this from the Storage card on the zPod detail page in zpodweb; see In zpodweb for what it looks like. All sizes are in GiB.

Permissions and limits¶

• Anyone with read access to the zPod can list disks, storages, folders and exports.
• A zPod OWNER or ADMIN can change them.
• A single disk can be at most zpodfactory_zpod_disk_max_gb (default 2048 GiB). Superadmins can exceed it.
• A disk add or grow is refused when it would leave the zcore VM's datastore with less free space than zpodfactory_datastore_min_free_pct (default 15%). The full size counts, even for a thin disk. This limit applies to everyone. See Settings & feature flags.
• Only one disk change can run at a time per zPod.

Quick start¶

To add a 500 GiB storage, export it, and mount it on the nested ESXi hosts:

❯ zcli zpod storage add zpod_name --size 500
❯ zcli zpod nfs add zpod_name STORAGE02 DS02
❯ zcli zpod nfs list zpod_name

Then mount zcore.<zpod domain>:/FILER/STORAGE02/DS02 as an NFS 3 datastore in the nested vCenter.

Disks¶

❯ zcli zpod disk list zpod_name
❯ zcli zpod disk add zpod_name --size 500
❯ zcli zpod disk grow zpod_name /dev/sdc --size 800
❯ zcli zpod disk remove zpod_name /dev/sdc

• list shows each disk's device, SCSI unit, size, and what zcore sees on it: system, protected, blank, foreign, or in-use with the storage name. It also shows the kind (zpod for the deploy's two disks, user for disks you added) and the status. --kind zpod,user filters the list, and -j gives JSON.
• add puts a blank thin disk on the next free SCSI unit of the zcore VM.
• grow sets a new, larger total size. You can grow the filer disk, but never the system disk. To use the new space, grow the storage on the disk afterwards (zcli zpod storage grow).
• remove deletes a blank or foreign disk you added, along with its data. It is refused while the disk is in use (409) and for the deploy's own disks (403). Add -y to skip the confirmation.

Disk changes run as zPod Engine flows (add disk <zpod>/<size>GiB by <user> and similar). By default each command follows its run and prints how it ended, with exit code 1 on failure. Pass --no-wait to return at once. A failed add leaves nothing behind. You can retry a failed grow or remove.

zcore identifies each disk by its serial, which is the vSphere disk UUID. A device name that changes after a reboot therefore changes nothing.

Storages¶

❯ zcli zpod storage list zpod_name

The list shows each storage with its usage, mount point, size, partition, layout and disk. Below each storage is its folder tree, with each folder's export, clients and access. The table caption shows the NFS server state. --kind zcore,user filters the list (zcore is STORAGE01), and -j gives JSON.

One-step commands¶

These commands combine the disk and storage steps:

❯ zcli zpod storage add zpod_name --size 500 [--lvm] [--name STORAGE05]
❯ zcli zpod storage grow zpod_name STORAGE02 --size 1000
❯ zcli zpod storage remove zpod_name STORAGE02

• add adds a new disk, then creates a storage on it. The default name is the next free STORAGEnn.
• grow with --size grows the disk to that total size first, then grows the storage into the new space, online.
• remove deletes the storage, then removes its disk with all its data. It asks for confirmation once (-y skips it).

Step by step¶

❯ zcli zpod storage create zpod_name --disk /dev/sdc [--lvm] [--name STORAGE02] [--dry-run]
❯ zcli zpod storage grow zpod_name STORAGE02
❯ zcli zpod storage delete zpod_name STORAGE02
❯ zcli zpod storage adopt zpod_name --device /dev/sdc1 [--name STORAGE02]

• create partitions, formats and mounts a blank disk of the zPod. The layout is a raw partition by default. Pass --lvm for a single LVM volume.
• grow without --size grows the storage into free space its disk already has, for example after zcli zpod disk grow. This works for STORAGE01 too.
• delete unmounts a storage. The filesystem and its data stay on the disk. It is refused while a folder of the storage is exported.
• adopt mounts an existing filesystem again as a storage, with its data intact. Pass a partition (/dev/sdc1) or an LVM volume (vg/lv).

A storage that still has an export is refused, whichever way you delete it:

create, adopt, grow and delete accept --dry-run, which shows the planned steps without running them, and --verbose, which shows each step's command and output. For example, growing an LVM storage:

❯ zcli zpod storage grow zpod_name STORAGE04 --dry-run --verbose

If a step fails, the CLI prints the error, the step that failed, and what was rolled back.

zcore only formats or adopts a disk that zPodFactory recorded for this zPod (403 otherwise). A disk attached to the zcore VM by hand is never touched.

Folders¶

❯ zcli zpod storage folder list zpod_name STORAGE02
❯ zcli zpod storage folder add zpod_name STORAGE02 DS02 [--owner root:root] [--mode 0777]
❯ zcli zpod storage folder set zpod_name STORAGE02 DS02 --mode 0755 [--owner user:group] [--recursive]
❯ zcli zpod storage folder remove zpod_name STORAGE02 DS02 [--force]

New folders default to owner root:root and mode 0777. set --recursive also applies the change to everything below the folder. remove refuses a folder that is exported. Pass --force to remove a folder that is not empty, along with its contents.

NFS exports¶

❯ zcli zpod nfs list zpod_name
❯ zcli zpod nfs status zpod_name

• list shows every export: the ones zcore's image writes to /etc/exports (owner system) and the ones you added.
• status shows the NFS server on one line, then each export with the hosts that currently have it mounted.

Add, change, remove¶

❯ zcli zpod nfs add zpod_name STORAGE02 DS02 [-c 10.42.111.0/24] [-o rw,no_subtree_check,no_root_squash]
❯ zcli zpod nfs set zpod_name STORAGE02 DS02 -c 10.42.111.0/24,10.96.1.10 [-o ro,no_subtree_check]
❯ zcli zpod nfs client add zpod_name STORAGE02 DS02 10.96.1.11
❯ zcli zpod nfs client remove zpod_name STORAGE02 DS02 10.96.1.11
❯ zcli zpod nfs remove zpod_name STORAGE02 DS02

• add exports a folder of a storage, and creates the folder when it is missing.
  • --clients/-c: who may mount it. Give an IPv4 address, an IPv4 network, or *, either comma-separated or as repeated options. The default is *, meaning any host that can reach zcore.
  • --options/-o: exports(5) options for every client. The default is rw,no_subtree_check,no_root_squash.
• set makes an export exactly what you give it. The client list is replaced, and so are the options when you pass -o. The export is created when it is missing.
• client add and client remove change one client. Removing the last client removes the export, so the CLI asks first.
• remove stops exporting the folder. The folder and its data always stay.

The list shows no_root_squash in orange as a reminder that root on the client is root on the export.

zcore's own exports are read-only through zPodFactory, and NFS-01 cannot be changed at all (403).

Mount an export¶

Mount an export in a nested vCenter as an NFS 3 datastore, with server zcore.<zpod domain> and the export path, for example /FILER/STORAGE02/DS02. On Linux:

mount -t nfs zcore.<zpod domain>:/FILER/STORAGE02/DS02 /mnt/ds02

zpodweb shows ready-to-copy ESXi and Linux mount commands next to each export.

zPod info¶

zcli zpod info <zpod_name> has a Storage panel, shown by default. It lists zcore's storages with size, usage and disk, and the NFS exports with their clients. Show it alone with -f s. On an older zcore, the panel says why it is empty.

In zpodweb¶

zpodweb 0.2.0 shows the same disks, storages and exports in a Storage card on the zPod detail page, under Networks. The screenshots below are from zpodweb 0.2.0 on a test zPod; click one to zoom in.

Storages tab¶

A line at the top gives the NFS server's state, its versions and threads, how many exports and mounted clients there are. Below it, each storage with its folders as a tree: a usage bar coloured by how full it is (the sizes on hover), the raw or LVM layout, the mount and path, the disk, and each folder's NFS clients and access. A pulsing dot marks a folder some host has mounted (who, on hover), and its share (zcore.<domain>:<path>) is one click from the clipboard. Protected items show a lock instead of a menu.

A storage's menu offers grow, add folder and delete; folders have their own menu for export, owner and mode, and removal:

Layout tab¶

Each of zcore's disks with what is on it: the layout, the storage with its usage and mount, and its folders with their NFS clients, plus the total space assigned for storage.

Dialogs¶

Add a storage	Export a folder over NFS
On a new disk, added through vCenter, or on a blank disk of the zPod; the dialog then follows the vCenter run and zcore's steps.	Any host or chosen clients (the zPod's networks offered in one click), read-write or read-only, no_root_squash, and ready-to-copy ESXi and Linux mount commands.

Delete a storage: keep the data on the disk (the default), or remove the disk too after typing the storage's name. An exported folder must be unexported first, and the dialog offers it in one click:

API¶

Disk changes answer 202 with a zPod Engine run.