Upgrading Vaultwarden to 1.37.x on Proxmox LXC: Security Patches and Bitwarden Client Fix (2026)
Home Lab

Upgrading Vaultwarden to 1.37.x on Proxmox LXC: Security Patches and Bitwarden Client Fix (2026)

Ricardo Gil
November 16, 2026
10 min read
#vaultwarden #proxmox #self-hosting #security #lxc

If you've been running Vaultwarden on Proxmox LXC and recently noticed your Bitwarden mobile or desktop apps throwing login errors, you're not alone — and there's a clear fix. On October 5, 2026, the Vaultwarden team released version 1.37.4, a security release that patches seven vulnerabilities, including one rated High (CVSS 8.1). More urgently, every release in the 1.37.x series introduced mandatory compatibility changes: running Vaultwarden 1.36.x or older guarantees that Bitwarden clients version 2026.7.0 and newer will refuse to connect. This guide walks you through how to upgrade your Vaultwarden LXC container safely, what to watch for in the breaking-change notes, and how to verify everything is working before you close the terminal. Sources: official Vaultwarden release notes on GitHub.

What Changed in Vaultwarden 1.37.x — and Why It Matters Now

The 1.37.x series shipped across four months and each point release brought something that affects operators running older installs.

1.37.0 (July 24, 2026) was the first security release of the cycle. Per the official release notes, it patched eight vulnerabilities ranging from SSRF via the icon endpoint (CVSS 5.8–6.3) to cross-organization cipher access (CVSS 5.9) and an unauthenticated WebSocket flooding denial-of-service (CVSS 5.3). Crucially, it also introduced a client compatibility boundary: Bitwarden clients version 2026.7.0 and later require this release or newer on the server side. If you're still on 1.36.x, those clients will fail silently at login.

1.37.2 (August 22, 2026) raised the bar again: clients version 2026.8.0 and later need 1.37.2 as a minimum. This is the release that broke self-hosters who kept auto-updating their Bitwarden apps but forgot about the server side.

1.37.3 (September 13, 2026) added rate limiting to the prelogin and auth-request endpoints, revoked 2FA remember tokens on credential changes, and fixed a password-change regression with newer web vaults. No new client minimum was set.

1.37.4 (October 5, 2026) is the version you should be running today. It patches seven security advisories filed with GitHub Security Advisories (GHSAs), the worst of which is a High-severity organization-member revocation flaw (GHSA-69q9-v8p6-xvx3, CVSS 8.1). The advisory details are still private pending CVE assignment, but the team's recommendation is unambiguous: update as soon as possible. Additional fixes cover 2FA bypass (CVSS 6.8), invitation spoofing (CVSS 6.8), attachment access (CVSS 6.5), organization event-log leaks (CVSS 4.3), cipher-sharing (CVSS 4.3), and an organization API key exposure (CVSS 3.8).

The table below summarizes the client compatibility matrix so you know exactly which server version you need:

Bitwarden client version Minimum Vaultwarden version Release date
2026.7.0 and later 1.37.0 July 24, 2026
2026.8.0 and later 1.37.2 August 22, 2026
Any current version 1.37.4 (recommended) October 5, 2026

If you're wondering how to tell what version Vaultwarden is running right now, navigate to the admin panel at https://your-vaultwarden-domain/admin (or the LXC IP directly on port 80 if you haven't set up a reverse proxy). The version string appears in the footer. If you set up Vaultwarden using the Proxmox LXC community script, it may also show in the LXC console if you run update or query the container's running image.

Before You Upgrade: Back Up the Data Directory

Your Vaultwarden data directory holds the SQLite (or MySQL/PostgreSQL) database, attachments, Sends, the ADMIN_TOKEN hash, and RSA key files. If the upgrade goes wrong or a migration step corrupts the database, you need that backup. I'd recommend you do this even if you have Proxmox Backup Server snapshots, because PBS snapshots are CT-level and restoring one would roll back the entire container, not just the data.

Log into the Vaultwarden LXC console from the Proxmox shell or directly with pct enter <CT_ID>, then find where data lives. For Docker Compose setups installed via the community scripts, the directory is typically /opt/vaultwarden/data. For Podman-based installs, it's often /vlt/. Run docker inspect vaultwarden | grep Mounts -A 20 (or podman inspect vaultwarden) to confirm before anything else.

bash
# Confirm where the data lives
docker inspect vaultwarden | grep -A 5 '"Type": "bind"'

# Back up the data directory (adjust path to match your setup)
DATA_DIR=/opt/vaultwarden/data
cp -a "$DATA_DIR" "${DATA_DIR}.bak.$(date +%Y%m%d)"
echo "Backup created: ${DATA_DIR}.bak.$(date +%Y%m%d)"

If your Vaultwarden uses a MySQL or MariaDB backend instead of the default SQLite, also dump the database before upgrading:

bash
mysqldump -u vaultwarden -p vaultwarden > /opt/vaultwarden/vw_backup_$(date +%Y%m%d).sql

Once the backup is confirmed, it's also worth taking a Proxmox CT snapshot from the Proxmox web UI: select your Vaultwarden LXC → Snapshots → Take Snapshot. This gives you a 30-second rollback if anything goes wrong. A reliable homelab UPS like the CyberPower CP1500PFCLCD UPS (~$220) ensures the host doesn't lose power mid-operation — SQLite corruption during an unexpected shutdown is exactly the kind of problem that makes you wish you'd done the backup.

Check price of CyberPower CP1500PFCLCD UPS on Amazon → https://www.amazon.com/dp/B00429N19W?tag=gilricardoweb-lab-20

Upgrading Vaultwarden on Proxmox LXC

How you upgrade depends on how the container was set up. The two most common methods on Proxmox are Docker Compose (via the community scripts) and Podman with a systemd unit.

Method A: Docker Compose (community scripts deployment)

If you used the Proxmox VE community scripts to set up Vaultwarden, the container is typically running Docker with a Compose file at /opt/vaultwarden/. The upgrade is a three-command sequence:

bash
# Navigate to the Compose directory
cd /opt/vaultwarden

# Pull the latest image from Docker Hub
docker compose pull

# Recreate the container with the new image (data volume stays intact)
docker compose up -d

# Confirm the new version is running
docker compose exec vaultwarden /vaultwarden --version 2>/dev/null || \
  docker logs vaultwarden 2>&1 | grep "Vaultwarden version"

docker compose pull downloads the new vaultwarden/server:latest image without touching the running container. docker compose up -d then recreates the container from the new image, re-mounting the same data volume. If you pinned an explicit version tag in docker-compose.yml (e.g., image: vaultwarden/server:1.36.0), update it to vaultwarden/server:1.37.4 or simply vaultwarden/server:latest before pulling.

Method B: Podman + systemd

Some Proxmox LXC setups run Vaultwarden under Podman with a systemd unit. The upgrade flow is similar — pull the new image and restart the unit:

bash
# Pull the latest image
podman pull docker.io/vaultwarden/server:latest

# Restart the systemd service to swap in the new image
systemctl restart vaultwarden.service

# Check it started cleanly
systemctl status vaultwarden.service

If you have an auto-update timer (a vaultwarden-update.timer systemd unit that checks GitHub releases), you can trigger it manually: systemctl start vaultwarden-update.service. The timer script compares the running image digest to the latest tag and replaces the container only if they differ.

Method C: Bare Docker without Compose

For installs that run a single docker run command without a Compose file, you'll need to stop the container, remove it (keeping the data volume), pull the new image, and re-run your original docker run command with the same flags:

bash
# Stop and remove the old container (the volume is NOT deleted by default)
docker stop vaultwarden
docker rm vaultwarden

# Pull the new image
docker pull vaultwarden/server:latest

# Re-run with your original flags (example — adapt to your config)
docker run -d \
  --name vaultwarden \
  --restart=unless-stopped \
  -v /opt/vaultwarden/data:/data \
  -p 80:80 \
  -e ADMIN_TOKEN="your_token_hash" \
  -e DOMAIN="https://vault.yourdomain.com" \
  vaultwarden/server:latest

The -v /opt/vaultwarden/data:/data mount is what preserves your vault between upgrades. As long as that path is consistent and you haven't used docker rm -v (which would delete the volume), your passwords are safe.

Post-Upgrade Verification

Once the container is running again, verify the upgrade worked before you open things up to clients. Check the container logs for the version string and any startup errors:

bash
# Docker Compose
docker compose logs --tail=30 vaultwarden

# Bare Docker or Podman
docker logs vaultwarden 2>&1 | tail -30

A clean startup looks like:

bash
INFO  vaultwarden::config - Reading config file '/data/config.json'
INFO  vaultwarden - Vaultwarden version v1.37.4 ...

Then open the Bitwarden web vault (your domain or the LXC IP) and confirm you can log in. If you have the admin panel configured, check https://vault.yourdomain.com/admin — it shows the running version and diagnostics. Test the mobile app as well, since it caches credentials differently from the browser extension.

If your Vaultwarden is behind a Traefik reverse proxy (see the guide on Traefik v3 on Proxmox LXC for setup details), you don't need to touch Traefik — the LXC IP and port haven't changed, only the container image. The TLS certificate, routing labels, and middleware rules are all defined in Traefik's config, not in the Vaultwarden image.

Breaking Changes to Address After Upgrading

The 1.37.x release notes list several operator-side changes that can silently break things if you don't address them. Here's what to review in your environment, per the official release notes:

Reverse proxy trusted IPs. If you use the IP_HEADER=X-Forwarded-For environment variable for rate limiting and IP-based access control, the behavior changed in 1.37.4. Vaultwarden now uses the rightmost address in the X-Forwarded-For chain that isn't listed in IP_HEADER_TRUSTED_PROXIES. If you have a CDN or multiple proxies in front (e.g., Cloudflare → Traefik → Vaultwarden), add every proxy layer to IP_HEADER_TRUSTED_PROXIES as a comma-separated list of CIDR ranges.

Removed experimental feature flags. If your Vaultwarden config (via environment variables or config.json) includes entries in EXPERIMENTAL_CLIENT_FEATURE_FLAGS that were removed in 1.37.4, Vaultwarden will start but log a warning, and saving settings in the admin panel will fail silently. Remove any obsolete flags from your config and restart.

Removed legacy API endpoints. POST /identity/accounts/register and POST /api/accounts/prelogin are gone. Per the release notes, no currently-shipped Bitwarden client uses them, so this only matters if you have custom scripts or integrations that call those endpoints directly.

DUO_USE_IFRAME deprecated. If you use Duo 2FA and had DUO_USE_IFRAME=true in your config, remove it — it's ignored now. The Duo integration otherwise works as before.

Organization API key rotation (recommended). The 1.37.4 release notes flag that before this update, organization admins could read the organization's API key. If you don't fully trust every admin of your Vaultwarden organization, rotate the API key after upgrading: Admin Console → Organizations → [your org] → Settings → Rotate API key.

If Bitwarden Clients Still Won't Connect

After a successful upgrade, most clients should reconnect within a session refresh or after a manual lock/unlock. If clients still fail after the upgrade, run through this checklist:

  • Check the client version. On mobile, go to Settings → About to see your Bitwarden version. If you're on 2026.7.0 or later and Vaultwarden is now on 1.37.4, the compatibility table says it should work. If clients are on an older version, update them too.
  • Clear the client cache. The Bitwarden browser extension caches server metadata. In Chrome/Firefox: Bitwarden extension → Settings → Security → Clear clipboard/cache, or uninstall and reinstall the extension entirely.
  • Check DOMAIN matches your reverse proxy URL. Vaultwarden's DOMAIN environment variable must exactly match the URL clients use to reach the server, including the scheme (https://) and any subdomain or path prefix. A mismatch can cause token validation to fail after the update.
  • Look at the container logs. Authentication failures show up with specific error codes. docker compose logs -f vaultwarden will show real-time login attempts and any rejection reasons.
  • Confirm TLS is working. If your Traefik certificate expired or the Let's Encrypt renewal failed, clients will reject the connection. curl -I https://vault.yourdomain.com from outside the LXC confirms the cert is valid.

If you need extra storage for Proxmox Backup Server snapshots so you can keep pre-upgrade restore points for longer, the WD_BLACK SN850X 1TB NVMe is a solid option for adding a fast backup drive to your Proxmox host. It's the same drive I'd look at for expanding storage on a Beelink GTi13 Ultra or Dell OptiPlex 7090.

Check price of WD_BLACK SN850X 1TB NVMe on Amazon → https://www.amazon.com/dp/B0B7CKVCCV?tag=gilricardoweb-lab-20

Keeping Vaultwarden Updated Going Forward

The pattern here — security release lands, clients update, old server breaks — is not new to Vaultwarden. The maintainers try to keep backward compatibility, but the 1.37.x cycle shows that security fixes sometimes require server-side API changes that old clients or old servers can't straddle. The safest approach is to subscribe to the Vaultwarden GitHub releases feed (there's an Atom feed at https://github.com/dani-garcia/vaultwarden/releases.atom) and treat security releases as "upgrade within the week" items.

For automation, Watchtower can poll Docker Hub for new images and restart the container automatically. The Vaultwarden maintainers recommend pinning to a specific version tag for production installs and reviewing release notes before auto-updating, since — as we saw with 1.37.0 — a point release can still carry config-breaking changes. A middle ground is to run Watchtower in monitor-only mode, which sends notifications when a new image is available without auto-applying it.

For the full picture on how the LXC container setup and Traefik reverse proxy fit together, the Proxmox LXC Complete Self-Hosting Guide covers the broader self-hosting stack. And if you're not already using Proxmox Backup Server for CT snapshots, this upgrade is a good excuse to get it configured — a 15-second rollback beats a 20-minute restore from a tar archive.

If your Proxmox host is running tight on memory and you want to give Vaultwarden, Traefik, and your other LXC containers more headroom without getting into swap-induced slowdowns during upgrades, browsing for compatible 32GB DDR4 SODIMM RAM on Amazon is worth the five minutes.

Want this kind of setup for your business?

I deploy private, self-hosted infrastructure on Proxmox: AI, automation and internal tools on hardware you own, documented and handed off. See the Private AI Stack →

Not sure what you need? Book a free discovery call.

Ricardo Gil is a full-stack developer (.NET/Angular) with 6+ years of experience. He runs a personal homelab on a Beelink mini PC with Proxmox VE, self-hosting everything from Jellyfin to local LLMs with Ollama. More about Ricardo →
📬Weekly Newsletter

Get the best home lab & AI content

No spam. One email per week. Unsubscribe anytime.

Share this article