Server

Introduction

A major Proxmox Backup Server upgrade changes several layers at once:

PBS 3.x → PBS 4.x
Debian 12 → Debian 13
older Proxmox kernel → newer Proxmox kernel

The system may boot successfully while still carrying:

  • repository residue;
  • failed services;
  • package inconsistencies;
  • stale configuration files;
  • datastore issues;
  • bootloader warnings;
  • mail compatibility warnings.

A structured read-only audit provides a safer way to validate the upgrade before attempting any cleanup.

1. Confirm the Version Stack

Start with:

hostnamectl
cat /etc/os-release
uname -a
proxmox-backup-manager versions

A fully migrated PBS 4 system should consistently show:

  • PBS 4.x;
  • Debian 13 Trixie;
  • a supported Proxmox kernel;
  • no running PBS 3 daemon.

Any mixture of Debian 12 and Debian 13 packages deserves further inspection.

2. Check System Health Before Configuration Details

Useful commands:

systemctl is-system-running
systemctl --failed
uptime
free -h
df -hT
df -ih

A healthy result typically includes:

systemctl is-system-running
running

and:

0 loaded units listed

from:

systemctl --failed

Also review:

  • memory pressure;
  • swap state;
  • root filesystem capacity;
  • inode exhaustion;
  • OOM history.

3. Validate PBS Services

The most important services include:

systemctl status proxmox-backup.service --no-pager
systemctl status proxmox-backup-proxy.service --no-pager

Both should normally be active.

A service that failed during boot but later recovered should not automatically be treated as a current fault.

Repeated restarts or persistent failed state are much more significant.

4. Verify the Datastore

List configured datastores:

proxmox-backup-manager datastore list

Then confirm the mount:

findmnt
df -hT

A datastore may be correctly configured in PBS while the underlying mount is missing.

Both layers must therefore be checked:

PBS datastore configuration
        +
actual filesystem mount
        =
usable datastore

5. Audit Backup Tasks

Use:

proxmox-backup-manager task list

Also inspect recent task results.

An active backup immediately after a reboot may produce:

  • high disk I/O;
  • elevated CPU usage;
  • increased network throughput;
  • hashing and compression load.

Those are expected conditions.

A completed task with OK is much more meaningful than a brief high load average.

6. Repository Consistency

Inspect repositories without refreshing them:

grep -Rhv '^[[:space:]]*#' \
  /etc/apt/sources.list \
  /etc/apt/sources.list.d/*.list \
  /etc/apt/sources.list.d/*.sources 2>/dev/null

Then inspect package policy:

apt-cache policy

A post-upgrade PBS 4 installation should not have active Bookworm/PBS 3 repositories mixed with Trixie/PBS 4 repositories.

Disabled historical repository files are harmless.

For example:

pbs-enterprise.sources.disabled
pbs-no-subscription.list.disabled

are not active package sources.

7. Check Package State

Use:

dpkg --audit

An empty result is desirable.

Also inspect PBS package versions:

proxmox-backup-manager versions

Upgrade residue such as:

*.dpkg-old
*.dpkg-dist
*.ucf-dist

is not automatically a problem.

Those files should be classified as:

  • harmless historical backup;
  • potentially active configuration conflict;
  • unresolved.

Deleting them only for cosmetic reasons provides little value.

8. Bootloader Warnings Require Context

Some Proxmox systems use GRUB directly rather than proxmox-boot-tool.

A command such as:

proxmox-boot-tool status

may report:

/etc/kernel/proxmox-boot-uuids does not exist

This does not necessarily indicate a boot problem.

If:

  • the machine boots successfully;
  • EFI entries are correct;
  • GRUB is installed correctly;
  • the upgrade checker accepts the bootloader state;

then the missing proxmox-boot-uuids file may simply mean that the installation uses a different supported boot path.

A warning should be interpreted according to the actual boot architecture.

9. ACPI and Firmware Noise

Kernel logs may contain messages such as:

ACPI BIOS Error
AE_NOT_FOUND
AE_ALREADY_EXISTS

If:

  • the messages occur only during boot;
  • services start normally;
  • storage remains healthy;
  • no kernel panic occurs;
  • no hardware instability follows;

then they usually belong in a firmware-warning category rather than an operational-error category.

Firmware noise should not be “fixed” solely to obtain a clean log.

10. Postfix Can Become the Only Remaining Upgrade Warning

A PBS upgrade to Debian 13 may also upgrade Postfix.

A common warning is:

support for parameter "smtp_use_tls" will be removed;
instead, specify "smtp_tls_security_level"

Another common warning is:

Postfix is using backwards-compatible default settings

These are not PBS service failures.

They belong to the local mail subsystem and should be audited separately.

11. Final PBS Health Criteria

A PBS 4 system can be considered healthy when:

  • PBS version is correct;
  • Debian release is correct;
  • supported kernel is running;
  • PBS services are active;
  • datastore is mounted and writable;
  • no failed systemd units remain;
  • repositories are consistent;
  • dpkg --audit is clean;
  • backup jobs complete successfully;
  • no active filesystem or I/O error exists;
  • no upgrade regression is visible.

A healthy system may still contain harmless upgrade residue or firmware messages.

Conclusion

A PBS major-version upgrade should not be considered complete merely because the web interface loads.

A better validation process checks:

  • package state;
  • service state;
  • datastore state;
  • repository state;
  • backup execution;
  • filesystem health;
  • kernel logs;
  • mail subsystem compatibility.

This approach provides a stronger operational conclusion than simply searching logs for the word error.

Leave a Reply

Your email address will not be published. Required fields are marked *