Troubleshooting Tips

This is part of Installing a Data Processor in Proxmox VE (Cluster).

  • VM hangs at "SeaBIOS ... Booting from Hard Disk...". The VM was created with the default legacy BIOS firmware. The Ubuntu 24.04 image requires UEFI. Recreate the VM with --bios ovmf (see Creating the Virtual Machines), or convert an existing VM: qm set <vmid> --bios ovmf then qm set <vmid> --efidisk0 vmstore:0,efitype=4m,pre-enrolled-keys=0, then start it.

  • Use the QCOW2 image, not the OVA. The OVA is VMware packaging. Importing it into Proxmox produces a VM configured with a vmxnet3 network adapter and a guest image whose initramfs assumes VMware storage drivers; the VM typically stops during early kernel initialization, immediately before mounting root, with one host CPU core spinning. The OVA import also requires a file-based working storage with the Disk image content type enabled, which ZFS pools cannot provide. The QCOW2 image avoids all of this.

  • The DAm console prompt displays DataProcessor(DR-master)>. After the DAm is converted to the DA-Master role, its console prompt renders as DR-master. DR stands for Data Refinery, a legacy term for the Data Analyzer that survives in the prompt string. This is cosmetic and does not indicate a misconfiguration — confirm the actual role in the DLm UI under System | DATA MANAGEMENT | Resources, where the Role column reports DA-master.

  • DNS is required for both the DL and the DA. The role configuration procedures include the required set interface management dns command.

  • VM appears stuck at first boot. With no DHCP server on the network, the Stellar Cyber VMs can take 10–15 minutes to come up. Close and re-open the console after 10 minutes. Once a static IP is configured, the delay no longer occurs.

  • Console access through a jump host. Forward the Proxmox web port and browse to the forwarded address:

    ssh -J <user>@<jumphost> -L 8006:<pve-host>:8006 root@<pve-host>

    Then open https://localhost:8006.

  • set cluster_interface rejects the name. Re-run show interface all and copy the interface name exactly as displayed. It is ens19 with the VM configuration in this document, but it varies with the virtual PCI topology.

  • Pool creation reports a device is in use. A partition table or filesystem signature survived the wipe. Re-run the wipe procedure for that disk. See Configure Storage.

  • zpool status shows sda/sdb instead of ata-* names. A raw device path was passed to zpool create. Destroy the pool and recreate it using /dev/disk/by-id/ paths.