backup connect failed: command error: failed to set owner on s3 backend

I

infraerik

New Member
Aug 4, 2026
5
1
3
I just setup a fresh install of PBS, following the documentation as best I understood it. I setup two datastores - one local directory and one S3.

Running the backups I keep seeing the following errors on certain VMs:

Code: 
ERROR: Backup of VM 102 failed - VM 102 qmp command 'backup' failed - backup connect failed: command error: failed to set owner on s3 backend

And when I check the contents of the backups up that datastore, the VM is missing.

Any hints on where to look to fix this?
 
The detail you need is on the PBS side. The actual S3 response will be logged there rather than the summarized message PVE gives you.

Could you check the task log for that backup in the PBS UI, or check the journal around the time it failed?

You can run:
Code: 
journalctl -u proxmox-backup-proxy --since "1 hour ago"
 
Hi,

> Running the backups, I keep seeing the following errors on certain VMs:

Does this mean some VM backups are working fine to the s3 datastore and some are not? Please also run the following S3 client sanity check and share the output:

Code: 
proxmox-backup-manager s3 check <s3-endpoint-id> <bucket>
 
Last edited:
JohnnieKF said:
The detail you need is on the PBS side. The actual S3 response will be logged there rather than the summarized message PVE gives you.

Could you check the task log for that backup in the PBS UI, or check the journal around the time it failed?

You can run:
Code: 
journalctl -u proxmox-backup-proxy --since "1 hour ago"
Click to expand...
I see a couple of entries like this:
Code: 
Aug 04 14:12:57 syn-pbs proxmox-backup-proxy[628]: Upload backup log to datastore 'ImpossibleFR', root namespace vm/100/2026-08-04T12:03:51Z/client.log.blob
Aug 04 14:14:41 syn-pbs proxmox-backup-proxy[628]: GET //api2/json/backup?backup-id=102&backup-time=1785845578&backup-type=vm&benchmark=false&debug=false&store=ImpossibleFR: 400 Bad Request: [client [::ffff:192.16>
 
infraerik said:
I see a couple of entries like this:
Code: 
Aug 04 14:12:57 syn-pbs proxmox-backup-proxy[628]: Upload backup log to datastore 'ImpossibleFR', root namespace vm/100/2026-08-04T12:03:51Z/client.log.blob
Aug 04 14:14:41 syn-pbs proxmox-backup-proxy[628]: GET //api2/json/backup?backup-id=102&backup-time=1785845578&backup-type=vm&benchmark=false&debug=false&store=ImpossibleFR: 400 Bad Request: [client [::ffff:192.16>
Click to expand...
Unfortunately some data's been missed off the end there:
400 Bad Request: [client [::ffff:192.16>...

Can you try grabbing this from the WebUI where the text should be wrapped? Or try:

journalctl -u proxmox-backup-proxy --since "1 hour ago" --no-pager
 
OK - with the --no-pager option for journalctl, I have the following:

Code: 
Aug 04 12:54:37 syn-pbs proxmox-backup-proxy[628]: error during snapshot file listing: 'unable to load blob '"/mnt/datastore/syn-pbs-repo01/vm/100/2026-08-04T10:52:09Z/index.json.blob"' - No such file or directory
(os error 2)'
Aug 04 13:08:37 syn-pbs proxmox-backup-proxy[628]: GET //api2/json/backup?backup-id=100&backup-time=1785841716&backup-type=vm&benchmark=false&debug=false&store=syn-pbs-repo01: 400 Bad Request: [client [::ffff:192.1
68.31.5]:37488] unable to acquire backup group lock "/run/proxmox-backup/locks/syn-pbs-repo01/vm-100" while creating locked backup group "/mnt/datastore/syn-pbs-repo01/vm/100"
Aug 04 13:09:48 syn-pbs proxmox-backup-proxy[628]: Upload backup log to datastore 'syn-pbs-repo01', root namespace vm/100/2026-08-04T10:52:09Z/client.log.blob
Aug 04 13:09:59 syn-pbs proxmox-backup-proxy[628]: rrd journal successfully committed (25 files in 0.013 seconds)
 
Please share the output of the following commands:
  • proxmox-backup-manager version --verbose
  • proxmox-backup-manager s3 check <s3-endpoint-id> <bucket> (as already requested by @shanv)
  • Retracted cat /etc/proxmox-backup/s3.cfg, making sure you do not leak any access and secret keys.
  • cat /etc/proxmox-backup/datastore.cfg
infraerik said:
Aug 04 13:08:37 syn-pbs proxmox-backup-proxy[628]: GET //api2/json/backup?backup-id=100&backup-time=1785841716&backup-type=vm&benchmark=false&debug=false&store=syn-pbs-repo01: 400 Bad Request: [client [::ffff:192.1 68.31.5]:37488] unable to acquire backup group lock "/run/proxmox-backup/locks/syn-pbs-repo01/vm-100" while creating locked backup group "/mnt/datastore/syn-pbs-repo01/vm/100"
Click to expand...
This error is a locking error and related to datastore syn-pbs-repo01 and might indicate that you have another concurrent operation acting on this group while trying to perform a backup (e.g. verification, prune, garbage collection). Check your task log, that might tell you more.

infraerik said:
failed to set owner on s3 backend
Click to expand...
This would indicate that there was an issue with setting the owner file on the s3 backend. Most likely either a connection issue or a configuration issue, but the s3 check command should help to identify that. Also, some errors you posted are related to the local datastore some to the s3 store afaict, so please provide a larger timespan of the systemd journal around the time of the backup failures to get the full picture.
 
OK - data dump as requested :)

Code: 
root@syn-pbs:~# proxmox-backup-manager version --verbose
proxmox-backup                    4.2.0         running kernel: 7.0.0-3-pve
proxmox-backup-server             4.2.4-1       running version: 4.2.0     
proxmox-kernel-helper             9.2.0                                   
proxmox-kernel-7.0                7.0.14-8                                 
proxmox-kernel-7.0.0-3-pve-signed 7.0.0-3                                 
ifupdown2                         3.3.0-1+pmx12                           
libjs-extjs                       7.0.0-5                                 
proxmox-backup-docs               4.2.4-1                                 
proxmox-backup-client             4.2.4-1                                 
proxmox-mail-forward              1.0.3                                   
proxmox-mini-journalreader        1.7                                     
proxmox-offline-mirror-helper     0.7.4                                   
proxmox-widget-toolkit            5.2.6                                   
pve-xtermjs                       6.0.0-2                                 
smartmontools                     7.5-pve2                                 
zfsutils-linux                    2.4.3-pve1

Code: 
proxmox-backup-manager s3 check <s3-endpoint-id> <bucket>
returns nothing at all - no message, no error

/etc/proxmox-backup/s3.cfg:
Code: 
s3-endpoint: ImpossibleFR
    access-key xxxxx
    endpoint {{bucket}}.{{region}}.storage.impossibleapi.net
    provider-quirks skip-if-none-match-header
    region eu-west-3
    secret-key xxxxx

As per the providers documentation: https://kb.impossiblecloud.com/en/h...e-as-a-datastore-in-proxmox-backup-server-pbs

Code: 
cat /etc/proxmox-backup/datastore.cfg
datastore: syn-pbs-repo01
    path /mnt/datastore/syn-pbs-repo01

datastore: ImpossibleFR
    backend bucket=pbs01,client=ImpossibleFR,type=s3
    comment
    gc-schedule daily
    notification-mode notification-system
    path /mnt/datastore/cache

Full journalctl log as attachment.
 

Attachments

infraerik said:
proxmox-backup-server 4.2.4-1 running version: 4.2.0
Click to expand...
As a first step, please upgrade to the latest version. Due to an unrelated bug [0], the package version on the left is currently not showing the installed one but rather the latest available one from the package repository. But the running version still shows you on 4.2.0. So please perform an
Code: 
apt update && apt full-upgrade
.

infraerik said:
proxmox-backup-manager s3 check <s3-endpoint-id> <bucket>returns nothing at all - no message, no error
Click to expand...
That is actually a good sign, so all the tests pass without issues. Also the config looks okay.

Therefore the question: is the `failed to set owner on s3 backend` error reproducible for VM 102? If so, please check and inspect the contents for the corresponding object on the S3 provider side. The object should be located at /ImpossibleFR/.cnt/vm/102/owner in this particular case.
Do you have object versioning or object locking enabled on this bucket? Note that these are currently not supported, see https://bugzilla.proxmox.com/show_bug.cgi?id=6780

[0] https://lore.proxmox.com/pbs-devel/20260727135940.267321-2-e.fastermann@proxmox.com/
 
OK - post-upgrade the backups pass correctly.

Object-Lock is set on the bucket since it must be applied at creation and I'm hoping PBS will end up adding support for it over time.
 
  • Like
Reactions: JohnnieKF
You must log in or register to reply here.
Top Bottom
Back