Vesta — Situation Guide

What to do when… — find your situation below and follow the commands. Every command is copy-pasteable; the only things you fill in are the links we email you.

This is the quick-reference card. For what any of it means — how licensing works, how to control a download, privacy and air-gap — see the Administrator Guide.

Shareable. Contains no secrets.

Filling in the placeholders

We email you links. Everywhere below, replace the angle-bracketed name with the matching link, keeping the quotes — the links contain characters your shell would otherwise interpret:

Placeholder What we send
<fingerprint link> one link, at first contact, before anything else exists
<deploy-kit link> the small tarball of scripts you run
<application link> the app bundle — ships with every release
<supporting-services link> large, changes rarely — first install or when it changes
<AI-models link> large, changes rarely — first install or when it changes
<license link> your machine-locked license.json

Links are time-limited. If one lapses, ask us for a fresh one — nothing you have already downloaded is lost, and the download resumes.

Which situation are you in?

Your situation Go to
Installing Vesta for the first time 1. First install
Re-installing on the same machine 2. Re-install
A new release is available 3. Update
My licence is expiring or has expired 4. Renew
I changed the GPU / moved to a new machine 5. Hardware change
I'm moving to a different Vesta edition 6. Edition change
The download is slow, stuck, or failing 7. Download problems
I'm locked out of the admin account 8. Locked out
Something is broken and I need help 9. Getting support

1. First install

You send us: your machine's hardware id. We send you: the fingerprint link first, then five links.

# ── Step 1: get the fingerprint helper and run it ──────────────────────────────
curl -fsSL "<fingerprint link>" -o vesta-fingerprint.sh
chmod +x vesta-fingerprint.sh
./vesta-fingerprint.sh

It prints one of these. Email us the whole output, then wait for your five links:

GPU_UUID=GPU-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx     # NVIDIA GPU server
HARDWARE_ID=XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX      # Apple Silicon Mac
# ── Step 2: get the deploy kit and move into it ───────────────────────────────
curl -fsSL "<deploy-kit link>" -o vesta-deploy-kit.tar.gz
mkdir -p vesta
tar xzf vesta-deploy-kit.tar.gz -C vesta
cd vesta

# ── Step 3: download the bundles (the large parts) ────────────────────────────
#    Pass every bundle link you were sent in ONE command.
./vesta-download.sh "<application link>" "<supporting-services link>" "<AI-models link>"

# ── Step 4: put your licence where Vesta reads it ─────────────────────────────
mkdir -p license
curl -fsSL "<license link>" -o license/license.json

# ── Step 5: install ───────────────────────────────────────────────────────────
./install.sh

install.sh prints the address and a temporary admin password. Log in, change it, and confirm under License in the portal.

Check your prerequisites first — Docker, a GPU driver (or Docker Desktop with Model Runner on a Mac), python3, and about 45 GB free. Full table: Administrator Guide §4.


2. Re-install (same machine)

Wiping the deploy directory, rebuilding the OS, or starting the install over. Your licence still applies — it is tied to the machine, not to an installation.

# If you still have the deploy kit, the bundles and license/license.json:
cd vesta
./install.sh

If you no longer have them, ask us to re-send your links, then follow First install from step 2 — your existing licence is re-issued unchanged.

This is not a data-recovery procedure. If you removed the Docker volumes (for example with docker compose down -v), your documents and knowledge graph are gone and re-installing gives you an empty system. See What Vesta cannot do for you yet.


3. Update to a new release

Updates are free while your licence is valid, and your data is preserved.

cd vesta

# ── Step 1: refresh the deploy kit FIRST — do not skip this ───────────────────
curl -fsSL "<deploy-kit link>" -o vesta-deploy-kit.tar.gz
tar xzf vesta-deploy-kit.tar.gz -C .

# ── Step 2: download whichever bundles we sent (usually just the application) ─
./vesta-download.sh "<application link>"

# ── Step 3: apply ─────────────────────────────────────────────────────────────
./update.sh

Why step 1 matters. Each deploy kit belongs to one release and carries the checksums for that release's bundles. A new bundle checked against an old kit is refused with no kit checksum matches …. Refreshing the kit keeps them in step — and it will not touch your .env or your license/ folder.


4. Renew a licence

Applies equally to renewing early, renewing during the grace period, renewing after expiry, and converting a trial to a full licence. Nothing is downloaded but the licence file, and your data is never touched.

cd vesta
curl -fsSL "<license link>" -o license/license.json
docker compose restart api worker

Confirm the new expiry date under License in the portal.

When What you see What happens
~60 / 30 days before expiry amber "renew soon" banner runs normally
after expiry, within the grace period loud "expired — renew now" still runs (grace is up to 14 days)
after the grace period stops starting — your data is untouched and returns the moment you renew

5. Hardware change

Replacing a GPU, replacing a Mac's logic board, or moving to a different machine. Your licence is locked to hardware, so it must be re-bound.

# On the NEW hardware, in your deploy directory (the helper is already in your kit):
cd vesta
./vesta-fingerprint.sh

Email us that output and your License ID (portal → License), tell us it's a hardware change, then:

curl -fsSL "<license link>" -o license/license.json
docker compose restart api worker

Moving to a different machine also means moving your data. Copy these Docker named volumes across as part of your normal server migration, then apply the re-bound licence:

vesta-postgres-data    vesta-neo4j-data    vesta-documents    vesta-license-state

6. Edition change

Moving between Vesta editions (for example to a larger model tier). This is not an update — the edition ships a different model set, so you receive a full set of links.

cd vesta

curl -fsSL "<deploy-kit link>" -o vesta-deploy-kit.tar.gz && tar xzf vesta-deploy-kit.tar.gz -C .
./vesta-download.sh "<application link>" "<supporting-services link>" "<AI-models link>"
curl -fsSL "<license link>" -o license/license.json
./update.sh

Your data carries over unchanged — documents, knowledge graph and search index are all preserved, with no re-indexing.


7. Download problems

All of these run from your deploy directory. Nothing already downloaded is ever lost — every command resumes.

# How far along am I?
./vesta-download.sh --status

# Is my connection the bottleneck? Downloads ~200 MB, writes nothing.
./vesta-download.sh --benchmark "<application link>"

# Stop cleanly at any time: press Ctrl-C. Resume by re-running the same command.
./vesta-download.sh "<application link>"

# Use fewer connections (or one), if your network prefers it:
./vesta-download.sh --parallel 1 "<application link>"

# Don't saturate the office line:
./vesta-download.sh --limit-rate 10M "<application link>"

# Start a bundle over from scratch:
./vesta-download.sh --restart "<application link>"
Message What it means
no kit checksum matches … Your kit and your link are from different releases — refresh the kit first, §3 step 1.
not enough free disk Free some space and re-run; anything downloaded already is kept.
the download link is expired or invalid Ask us for a fresh link, then re-run — it resumes.
Benchmark shows little gain from parallel Your own line or a corporate proxy is the limit. Send us the numbers.

Running it unattended (nohup, screen, a scheduled job) is fine — it detects it isn't on a terminal and prints one plain progress line every 30 seconds instead of animating.


8. Locked out of the admin account

cd vesta
./vesta-reset-admin.sh

Prints a new temporary password and forces a change at next login. Only the password changes — no data is touched. To choose the password yourself:

VESTA_NEW_ADMIN_PASSWORD='YourTemp-Pw1' ./vesta-reset-admin.sh

9. Getting support

Vesta never contacts us, so we cannot see your system. Send us a diagnostic bundle — it is redacted and contains no secrets:

cd vesta
./support-bundle.sh

It collects versions, licence status, health, container logs and machine facts into a single archive. Email that archive with a description of what you were doing.


Edition-specific notes

Everything above is the same on every edition, except:

NVIDIA GPU server Apple Silicon Mac
Fingerprint prints GPU_UUID=GPU-… HARDWARE_ID=…
You need Docker Engine + Compose v2, NVIDIA driver, python3 Docker Desktop with Model Runner enabled, Xcode Command Line Tools (xcode-select --install), 24–32 GB memory or more
During install.sh asks for sudo once, to record the machine id
Re-bind needed when the GPU is replaced the logic board is replaced

What Vesta cannot do for you yet

Stated plainly so you can plan around it:

  • There is no built-in backup tool. Your data lives in the Docker named volumes listed in §5. Include them in your normal server backup routine — and please talk to us before you need them, not after.
  • Downgrading to an earlier release is not supported. Database upgrades only move forward. If a release causes a problem, contact us: the fix is a corrected release, not a rollback.

Full detail on any step: Vesta Administrator Guide. Questions or a renewal: contact Ingenium AI Solutions, quoting the License ID from your portal's License page.