System Updates

Pulse ships as signed update bundles produced by the vendor. The Update System screen in the appliance TUI applies one to the running appliance. The wizard is the same whichever way the bundle reaches the appliance — only the first step, the source, differs.

This page is the full walkthrough. For database backups and rolling back to a previous version, see Operations.

What an update bundle is

Every bundle is signed by Pulse. The appliance verifies the signature chain before it shows you anything, and refuses to proceed if the bundle is unsigned, tampered with, or signed by a key that does not match the public key baked into the running system. There is no way to apply an unsigned bundle from the TUI — that is the whole point of the signing chain.

A valid bundle carries a manifest with the version string, build timestamp, a short changelog, and a flag for whether the new version requires a reboot. You see all of this on the Update Available screen before you commit.

Choosing a delivery path

There are three ways to get a bundle onto the appliance. Pick by how the appliance is deployed:

Deployment	Recommended path	Why
Virtualized (VMware ESXi/vSphere, Proxmox, Hyper-V, …)	CD/DVD-ROM	Attach the update ISO to the VM's virtual optical drive — the same gesture you used to install. No USB passthrough to set up.
Physical / bare-metal appliance	USB drive	No hypervisor, so hand-carried removable media is the simplest path.
Remote / unattended, or staged by support	SFTP upload	Support delivers the bundle over the network; nobody has to walk to the rack.

All three feed into the same wizard and the same safety net described below — the table is about convenience, not capability.

How an update is applied

Once you confirm, the appliance runs a fixed sequence. It is worth knowing what happens, because most of it is automatic safety:

1. Verify — re-checks the signature and checksum.
2. Back up the database — takes a fresh PostgreSQL backup before touching anything, so the pre-update state is always recoverable.
3. Import and switch — unpacks the new system and switches the boot profile to it.
4. Health check — waits up to 60 seconds for Pulse to answer on its health endpoint.
5. Auto-rollback on failure — if the switch or the health check fails, the appliance rolls itself back to the previous version automatically and reports the failure. You are left on a working system, not a broken one.
6. Clean up — keeps the five most recent system generations and prunes older ones.

If the manifest says a reboot is required (typically a kernel change), the final screen tells you so; reboot from the main menu when convenient.

TIP

Because the appliance backs up the database and self-rolls-back on a failed health check, a bad update is largely self-correcting. The manual recovery tools on the Operations page are for problems that surface after the update is already running.

Before you update

A short checklist keeps an update boring:

• Confirm the appliance is healthy now — the web interface loads and the system is behaving. Updating a sick system makes diagnosis harder.
• Make sure you can reach the console — a local monitor and keyboard, or SSH as the pulse user. You will want to watch progress and reboot if asked.
• Plan for a possible reboot — some updates need one. Pick a window where a short restart is acceptable.
• Have the bundle ready in the right place for your chosen path (see each path below).
• (Optional) Take a manual backup from the Backup / Restore screen. The update takes one automatically, but an extra known-good snapshot is cheap insurance before a big jump.

Delivery paths in detail

Virtualized appliance — CD/DVD-ROM

This is the recommended path for any appliance running as a VM.

1. Copy the update ISO (pulse-update-<version>.iso) to your hypervisor's datastore.
2. Edit the VM's settings and point its CD/DVD drive at that ISO. In VMware vSphere/ESXi, set the drive to Datastore ISO File, browse to the ISO, and — this is the step people miss — tick Connected (and Connect At Power On). Other hypervisors have the same two ideas: choose the ISO, and make sure the drive is actually connected.
3. On the appliance, open Update System → CD/DVD-ROM. The TUI mounts the virtual drive, finds the bundle, and verifies it.

WARNING

If the TUI reports no disc, the drive is almost always attached but not Connected. Detection keys off the disc being mounted, not merely present in the VM's configuration. Tick Connected and retry.

Do not disconnect the disc while the update is running — the Updating... screen warns about this in yellow. The drive is unmounted automatically when the update finishes.

Physical appliance — USB drive

1. Format a USB stick as FAT32 or ext4.
2. Copy the bundle files to the root of the drive (a single top-level folder such as updates/ also works). Do not bury them deeper.
3. Plug it into the appliance and open Update System → USB Drive. The TUI scans for removable drives:
   • One drive found → it mounts and scans automatically.
   • More than one → the Select USB Device step lets you pick the right one.
   • None → it reports no USB drive found (format must be FAT32 or ext4) and waits.

Do not unplug the stick while the update is running — the bundle is read from it as it is applied, and the Updating... screen warns in yellow. It is unmounted automatically on completion.

Remote / support-staged — SFTP upload

1. The party delivering the bundle must already be an authorised SFTP user — see SFTP Access to authorise them, either by their SSH key or with the appliance's SFTP password.
2. They upload the bundle into the exchange directory under uploads/updates/.
3. On the appliance, open Update System → SFTP Upload. The TUI scans the exchange directory and verifies what it finds.

A successfully applied SFTP bundle is removed from the exchange directory afterwards, so it cannot be applied twice.

The wizard, step by step

Whichever source you pick, the rest is identical:

1. Source — pick USB Drive, SFTP Upload, or CD/DVD-ROM. The TUI scans, mounts (USB and CD/DVD-ROM), and verifies the bundle.
2. Manifest review — the Update Available screen shows the version, build timestamp, changelog, and whether a reboot is required. Read it. Press Esc to back out cleanly without applying anything.
3. Confirm — press Enter on the manifest screen to start. There is no second yes/no prompt; the manifest screen is the confirmation.
4. Progress — the Updating... screen streams progress as the bundle is verified, backed up, imported, and activated. If removable media is still mounted, it reminds you in yellow not to remove it.
5. Done — the final screen shows Update Successful! (green) or Update Failed (red) with the underlying error. Press Esc to return to the main menu.

After the update

• If the manifest flagged a reboot, reboot from the main menu's Reboot System entry.
• Confirm the web interface comes back and the version is the one you expected.
• Once you are satisfied, take a fresh backup from Backup / Restore so your next known-good snapshot is post-update.

Troubleshooting

What you see	Likely cause	What to do
no USB drive found (format must be FAT32 or ext4)	Stick isn't FAT32/ext4, or wasn't detected	Reformat as FAT32 or ext4, reseat the stick, retry
USB mounts but no bundle is found	Bundle isn't at the drive root, or files are incomplete	Put manifest.json + system.tar (and the .sha256 / .sig files) at the root or in a single top-level folder (e.g. updates/) — do not nest deeper
CD/DVD-ROM: no disc detected	ISO not attached, or attached but not Connected	Attach the ISO, tick Connected / Connect At Power On, retry
SFTP: no bundle found	Not uploaded to uploads/updates/, or sender not authorised	Confirm the upload path; authorise the sender (SSH key or password) via SFTP Access
Verification / signature error	Bundle is unsigned, corrupted, or signed with the wrong key	Get a correctly signed bundle from Pulse; re-applying the same file won't help
Update Failed during apply	Switch or health check failed — the appliance already auto-rolled-back	You're back on the previous version. Capture the error and contact support with it
Web app unreachable after a reboot	New version is unhealthy	Roll back to the previous generation, then restore a backup if needed

When an update goes wrong

A failed update auto-rolls-back during apply (see the troubleshooting table above). For problems that surface after the update is running, use Rollback to return to the previous system generation. Rollback is system-level only — it never touches the database or other data under /var. See Operations for backups, restore, and rollback in full.