Troubleshooting

Each section starts with the message as the scripts print it, so you can search this page for what is on your screen. For the meaning of a single exit code see Exit codes.

First: read the error and find the logCopy Anchor Link

A failing command produces one line of this form, followed by the output of that command:

in line 24: exit code 100 (APT: Package manager error (broken packages / dependency problems)): while executing command apt install -y build-essential

When the failure happened inside the container, the host adds Installation failed in container <CTID> (exit code: <n>), the path of a combined log, and a menu that asks what to do with the container. The menu and its options are explained in Container creation.

Where the logs areCopy Anchor Link

SESSION_ID is a random eight-character ID for the run. With dev_mode=logs the host logs are written to /var/log/community-scripts/ instead of /tmp.

To read the install log of a container that still exists:

pct exec 105 -- sh -c 'tail -n 60 /root/.install-*.log'

See more next timeCopy Anchor Link

Run the script again in verbose mode to see every command's output live. Choose Advanced Install and answer the verbose question with yes, or:

var_verbose=yes bash -c "$(curl -fsSL https://raw.githubusercontent.com/community-scripts/ProxmoxVE/main/ct/gotify.sh)"

dev_mode=keep keeps a failed container without asking, see Dev mode.

DNS: "Git DNS"Copy Anchor Link

✖️  Git DNS:github.com:(  ✖️  )raw.githubusercontent.com:(  ✖️  )api.github.com:(  ✖️  )
✖️  Interrupted by user (SIGINT)
✖️  Installation failed in container 105 (exit code: 130)

The install script checks at its start whether the container can resolve github.com, raw.githubusercontent.com and api.github.com. If one of them fails, the install stops right there. A name that resolves is shown with a check mark instead. The following Interrupted by user (SIGINT) and exit code 130 are how that stop is carried out; nobody pressed Ctrl-C.

The container has an address but its DNS server does not answer, or it answers without these names.

• Check which DNS server the container uses: pct exec <CTID> -- cat /etc/resolv.conf. Without a setting of its own a container uses the host's DNS configuration (node → System → DNS).

• Give the container a working DNS server. In Advanced Install it is the DNS server step, on the command line:

  Copy Text

  var_ns=1.1.1.1 bash -c "$(curl -fsSL https://raw.githubusercontent.com/community-scripts/ProxmoxVE/main/ct/gotify.sh)"

• A filtering resolver (Pi-hole, AdGuard Home) must not block the three names.

• With Tailscale on the host, MagicDNS is a known cause. Disable it temporarily.

Related messages with the same cause:

When an install fails with a network error, the menu offers Retry with DNS override in LXC (8.8.8.8 / 1.1.1.1). It rebuilds the container and writes those two servers into it before the install.

No networkCopy Anchor Link

No addressCopy Anchor Link

No IP assigned to CT 105 after 60 attempts

Exit code 118. The container started but got no IPv4 or IPv6 address on eth0 within about two minutes.

• DHCP: is a DHCP server reachable on the chosen bridge and VLAN? Test with another container or VM on the same bridge.
• Static address: check address, prefix and gateway. A gateway outside the subnet is dropped with Ignoring var_gateway '<gw>' because it is not in subnet of var_net '<net>'.
• Bridge and VLAN: the default bridge is vmbr0. Check that the bridge is connected to the network you expect and that a VLAN tag, if set, exists there.
• Firewall: a Proxmox VE firewall rule that drops DHCP.

Network bridge 'vmbr1' does not exist on this node

Exit code 208, before anything is created. The message lists the bridges that do exist. Pick one in Advanced Install or set var_brg.

Address but no internetCopy Anchor Link

Check the gateway of the container, the routing and NAT of your network, and outgoing firewall rules. A container on an isolated or internal bridge has no route out. If the host needs a proxy, set var_http_proxy; it is written into the container before packages are installed.

Storage content types (exit codes 119 and 120)Copy Anchor Link

No storage on this host is enabled for container templates (content 'vztmpl').

No storage on this host is enabled for container disks (content 'rootdir').

Exit code 120 for templates, 119 for disks. Every Proxmox VE storage has a list of content types it accepts. A container needs one storage that accepts templates and one that accepts container disks. On a default installation that is local and local-lvm; the error appears when the content type was switched off or the storage layout was changed.

When interactive, the script offers a dialog No storage for container templates (or container disks) that lists the storages able to hold the content and enables it on the one you choose, keeping what the storage already holds.

By hand, in the web interface: Datacenter → Storage → select the storage → Edit → Content. On the command line, list the current content first and then set the full list including the new type:

pvesm status
grep -A6 '^dir: local' /etc/pve/storage.cfg
pvesm set local --content backup,iso,vztmpl

pvesm set --content replaces the list, so repeat the types that are already there. Templates need a file-based storage (dir, nfs, cifs, btrfs, cephfs). Container disks can also be on lvm, lvmthin, zfspool or rbd.

Related messages:

Template downloadCopy Anchor Link

Template download failed after 3 attempts: debian-13-standard_13.1-1_amd64.tar.zst

Exit code 222. The line after it names the cause the script could determine.

To test by hand:

pveam update
pveam available -section system | grep debian-
pveam download local debian-13-standard_13.1-1_amd64.tar.zst

Related messages:

How templates are chosen, including the linuxcontainers.org fallback, is described in Container creation.

Unsupported Proxmox VE versionCopy Anchor Link

This version of Proxmox VE is not supported.
Supported: Proxmox VE version 8.0 – 8.9

This version of Proxmox VE is not yet supported.
Supported: Proxmox VE version 9.0 – 9.2

The container scripts run on Proxmox VE 8.0 to 8.9 and 9.0 to 9.2 and exit with code 1 on anything else. Proxmox VE 7 is too old: upgrade the host. A version newer than the range was released after the engine was last adjusted: look for an open issue in the ProxmoxVE repository.

A different message means the host is in the supported range but one package is behind:

pve-container <version> on this host cannot create Debian 13 containers.

Exit code 231. The pve-container package is older than the OS release you want to create. Update the host:

apt update && apt dist-upgrade

If apt finds nothing newer, the host has no working Proxmox VE package repository. The post-pve-install tool can set one up. To continue without updating, create the container with the older release the message suggests, for example var_version=12.

Container creation failed due to broken Perl modules on the PVE host.

Exit code 232. A partial upgrade left the Proxmox tools in a broken state. Run apt update && apt dist-upgrade -y, reboot, and try again.

Out of memory during a build (exit codes 137, 134, 243)Copy Anchor Link

exit code 137 (Killed (SIGKILL / Out of memory?))
exit code 134 (Process aborted (SIGABRT - possibly Node.js heap overflow))
exit code 243 (Node.js: Out of memory (JavaScript heap out of memory))

A compiler or a JavaScript build needed more memory than the container has. Builds of Node.js, Rust and Go applications are the usual cases.

During an install the menu offers

Retry with more resources (RAM: 2048→4096 MiB, CPU: 2→4 cores)

which rebuilds the container with doubled memory and cores. It can be chosen twice; after that the menu shows OOM retry exhausted (already retried 2x). To start with more memory right away:

var_ram=8192 var_cpu=4 bash -c "$(curl -fsSL https://raw.githubusercontent.com/community-scripts/ProxmoxVE/main/ct/fluid-calendar.sh)"

You can lower the memory in the container's Resources tab once the install is done.

During an update the same build runs again in the existing container. Raise its memory on the host, then update:

pct set 105 -memory 4096
pct exec 105 -- update

If the container has less memory or fewer cores than the script declares, the update warns before it starts:

Required: 2 CPU, 4096MB RAM | Current: 1 CPU, 1024MB RAM

Take that warning seriously for applications that compile on update.

For Node.js builds the error handler also prints

Possible Node.js heap OOM. Try: export NODE_OPTIONS="--max-old-space-size=3072" and rerun the build.

The number is 75 % of the container's memory, between 1024 and 12288. This helps when the container has enough memory but Node's own heap limit is lower.

Exit code 137 is also what you see when the host itself ran out of memory and its kernel killed the process. Check dmesg | grep -i oom on the host.

GitHub rate limitCopy Anchor Link

GitHub API rate limit, waiting 2s... (attempt 1/3)
GitHub API rate limit exceeded (HTTP 403).
To increase the limit, export a GitHub token before running the script:
  export GITHUB_TOKEN="ghp_your_token_here"

GitHub allows a limited number of unauthenticated API requests per hour per IP address. Several installs or updates in a row, or a shared address (carrier-grade NAT, VPN), use it up.

In many cases the scripts work around the limit on their own and only warn:

GitHub API rate limit reached; read the release from github.com instead. A token avoids this: var_github_token

Otherwise:

• Wait an hour, or

• use a token. Create one at https://github.com/settings/tokens; it needs no permissions beyond reading public repositories. Then either answer the prompt Would you like to enter a GitHub Personal Access Token (PAT)? [y/N]: with yes, or pass it in:

  Copy Text

  var_github_token=ghp_xxxxxxxx bash -c "$(curl -fsSL https://raw.githubusercontent.com/community-scripts/ProxmoxVE/main/ct/gotify.sh)"

  For updates inside a container: var_github_token=ghp_xxxxxxxx update.

  To set it once for all installs, put var_github_token=ghp_xxxxxxxx into /usr/local/community-scripts/default.vars. That file is only read in the User Defaults mode. See Defaults.

GitHub API authentication failed (HTTP 401). together with Your GITHUB_TOKEN appears to be invalid or expired. means the token itself is the problem.

Cluster filesystem and quorumCopy Anchor Link

The Proxmox cluster filesystem is not mounted at /etc/pve (pve-cluster: failed).

Exit code 228. Every pct, qm and pvesm command needs /etc/pve, which the pve-cluster service provides. This also applies to a single node without a cluster. When interactive, the script offers to restart the service. By hand:

systemctl restart pve-cluster
journalctl -u pve-cluster -n 50

/etc/pve is mounted, but /etc/pve/storage.cfg is missing.

Exit code 228. The cluster configuration is damaged. Check pvecm status and the journal of pve-cluster.

This cluster node has no quorum, so /etc/pve is read-only and nothing can be created.

or, later in the run, Cluster is not quorate. Start all nodes or configure quorum device (QDevice). Exit code 210. In a cluster, a node may only change configuration while it sees a majority of the nodes. Bring the other nodes back online and check:

pvecm status

A two-node cluster loses quorum as soon as one node is off; that is what a QDevice is for.

Container ID 105 not listed in 'pct list' after 10s.

Exit code 215. The container was created but the cluster filesystem had not caught up. Check with pct list a moment later; the container usually exists.

The console was closed during an installCopy Anchor Link

Closing the browser tab, losing the SSH connection or closing a laptop lid does not abort an install. The install keeps running inside the container, and the host process keeps waiting for it, with its output going to the build log instead of the terminal. Verbose runs say so up front:

Closing this console does not abort the install; it finishes on its own (Ctrl-C does abort).
Follow it with: pct exec 105 -- tail -f /root/.install-1a2b3c4d.log

To see how it went, open a new shell on the host:

pct exec 105 -- sh -c 'tail -n 30 /root/.install-*.log'
grep -h 'terminal closed' -A20 /tmp/create-lxc-*.log

The build log contains The terminal closed; continuing, output follows here. from the moment of the disconnect. Possible endings:

If you are not sure whether a container from such a run is complete, look for the last lines of its install log. A finished install ends with the cleanup step (Cleaned). When in doubt, remove the container and run the script again.

Ctrl-C is not the same as closing the console: it stops the container and ends the run with exit code 130.

UpdatesCopy Anchor Link

Still stuckCopy Anchor Link

• Look up the exit code in Exit codes.
• Search the issues of the ProxmoxVE repository for the application name; a broken upstream release usually has one already.
• When you open an issue, include the full error output, the Proxmox VE version (pveversion), whether you used default or advanced settings, and the relevant part of the combined log. Check the log for passwords and tokens before posting it.