Backup status "ok" but backup is not

[oliwel]

Hi,

today I stumbled by chance over a very uncomfortable behavior of the backup process. I had two containers failing but the backup report shows "ok".

First, I went into the same issue as https://forum.proxmox.com/threads/pbs-backup-stack-overflow.150658/ - reason was also a very deep folder structure created by a broken Wordpress. In the end I get a single line WARN in the logs but the overall status says "ok" with a backup size of 0 bytes...

4135: 2026-05-15 09:25:25 INFO: processed 77.444 GiB in 5m, uploaded 1.468 GiB
4135: 2026-05-15 09:25:25 INFO: thread 'tokio-runtime-worker' (854098) has overflowed its stack
4135: 2026-05-15 09:25:25 INFO: fatal runtime error: stack overflow, aborting
4135: 2026-05-15 09:25:25 INFO: adding notes to backup
4135: 2026-05-15 09:25:25 WARN: unable to add notes - proxmox-backup-client failed: Error: unable to update manifest blob - unable to load blob '"/mnt/datastore/.../ct/4135/2026-05-15T07:20:24Z/index.json.blob"' - No such file or directory (os error 2)
4135: 2026-05-15 09:25:25 INFO: cleanup temporary 'vzdump' snapshot
4135: 2026-05-15 09:25:25 INFO: Finished Backup of VM 4135 (00:05:01)

Even worse, I have an unpriviledged container where some files are owned by nobody with 700 permission, the backup process from the host fails to read them

INFO: failed to open file: "js": access denied
INFO: failed to open file: "php-compat": access denied

The job continues and reports a successful backup.

In both situations I would expect some kind of error shown in the report and on the backup list.

best regards

Oliver

 

[Chris]

Hi,

Hi,

today I stumbled by chance over a very uncomfortable behavior of the backup process. I had two containers failing but the backup report shows "ok".

First, I went into the same issue as https://forum.proxmox.com/threads/pbs-backup-stack-overflow.150658/ - reason was also a very deep folder structure created by a broken Wordpress. In the end I get a single line WARN in the logs but the overall status says "ok" with a backup size of 0 bytes...

4135: 2026-05-15 09:25:25 INFO: processed 77.444 GiB in 5m, uploaded 1.468 GiB
4135: 2026-05-15 09:25:25 INFO: thread 'tokio-runtime-worker' (854098) has overflowed its stack
4135: 2026-05-15 09:25:25 INFO: fatal runtime error: stack overflow, aborting
4135: 2026-05-15 09:25:25 INFO: adding notes to backup
4135: 2026-05-15 09:25:25 WARN: unable to add notes - proxmox-backup-client failed: Error: unable to update manifest blob - unable to load blob '"/mnt/datastore/.../ct/4135/2026-05-15T07:20:24Z/index.json.blob"' - No such file or directory (os error 2)
4135: 2026-05-15 09:25:25 INFO: cleanup temporary 'vzdump' snapshot
4135: 2026-05-15 09:25:25 INFO: Finished Backup of VM 4135 (00:05:01)

As you already discovered, this is an inherent issue with the current implementation of the proxmox-backup-client, as its stack grows beyond the limits set by the runtime environment. The fix is however non-trivial, requiring some in-depth changes. This is tracked by issue https://bugzilla.proxmox.com/show_bug.cgi?id=3001

Even worse, I have an unpriviledged container where some files are owned by nobody with 700 permission, the backup process from the host fails to read them

INFO: failed to open file: "js": access denied
INFO: failed to open file: "php-compat": access denied

These are treated as non-fatal on purpose, but I see your concerns.

The job continues and reports a successful backup.

In both situations I would expect some kind of error shown in the report and on the backup list.

Could you open an enhancement request for the improved error/warning reporting. This needs some discussion on a mechanism to be used for communicating these to the caller first.