Install on Linux

Install on Linux

Docveta runs on any Linux distribution with systemd, on PCs, servers and boards (Raspberry Pi 4/5, Rockchip, Radxa Cubie A7A, Orange Pi, Ampere). You don't need to install a database first.

Install (one command)

curl -fsSLO https://github.com/anand34577/docveta/releases/latest/download/get-docveta.sh
sudo sh get-docveta.sh

The script:

• downloads the latest Docveta for your processor and checks every file against the release's checksums
• installs it as the service docveta (program in /opt/docveta, data in /var/lib/docveta)
• adds text recognition for this machine: the NPU on Allwinner A733 boards (Radxa Cubie A7A), otherwise the processor (x64 and ARM 64-bit). Docveta starts and connects the engine itself; there are no tokens to copy.

Then open the address it prints, click Save and start to use the built-in database, and create your account. From another computer the page asks for a setup code:

sudo cat /var/lib/docveta/setup-code.txt

Upgrade: run the same two lines again. Your data and settings are kept. The script downloads and checks everything first, stops Docveta only while it swaps the files, updates the database on the first start, and waits until Docveta answers. If anything fails, Docveta is started again; if the new NPU engine fails its check, the previous one stays.

Remove: sudo sh get-docveta.sh --uninstall. The program and the NPU engine are removed; your documents and settings stay in /var/lib/docveta (installing again picks them up). To delete them too: sudo rm -rf /var/lib/docveta.

Option
--port 9000	Use another port (default 8080)
--version 0.4.0	Install a specific version
--no-ocr	Skip text recognition (born-digital PDFs are still searchable)
--uninstall	Remove Docveta (data is kept)

Everyday commands:

sudo systemctl status docveta
sudo systemctl restart docveta       # after editing /opt/docveta/docveta.conf
journalctl -u docveta -f             # live log

Proxmox containers (LXC)

On the Proxmox host, give the container the NPU (if the board has one) and the fast cores, then run the normal install inside the container:

curl -fsSLO https://github.com/anand34577/docveta/releases/latest/download/get-docveta.sh
sudo sh get-docveta.sh --proxmox <CTID>

By hand

Download docveta-<version>-linux-<arch>.tar.gz (uname -m: x86_64 → x64, aarch64 → arm64, armv7l → armv7), unpack it, optionally unpack docveta-ocr-<version>-linux-<arch>.zip into the same folder (it creates ocr/), and run sudo ./install.sh. Or just run ./docveta: it starts on port 8080 and keeps its data in ./data.

Don't run Docveta as root. Use a normal user or the service; the built-in database refuses to run as root.

Boards and small servers

• Keep Docveta's data folder on an SSD, NVMe or eMMC, not an SD card: SD cards are slow for databases and wear out.
• 2 GB of RAM is enough for Docveta and its database; text recognition needs about 1 GB more while it works.