Proxmox Backup Task Logs missing for single backup job

[Grus]

I currently have two backup jobs running.
Three PVE nodes in a cluster, with a single PBS as target.

Job 1: Daily runs
Job 2: Weekly runs



VM's included in job 1 has each backup run listed in the task history for each VM. Everything is good.

VM's included in job 2 has NO backup entries in the task history for each VM.
The VM below has been backed up two times, which is validated by looking at the backup storage.



How can this be? Is there a setting that I am missing?

 

[dcsapak]

was the backup run at all? i.e. do you see the corresponding backups on the pbs side? is there a 'global' task log for the job ?

 

[Grus]

The backups are on PBS. The times match. I did a manual run around 20:30, and the automated schedule starts at 02:00 and continues for a handfull of hours.


Looking at journalctl on the PVE node with the following search terms, reveals that backups was run indeed.


After doing some further searching in journalctl I also found this of interest on one of the PVE nodes. My PBS is on 10.0.0.8, and "pbs01-backup01-weekly" is the name of the mounted datastore (in PVE, it has another name in PBS) in use for that backup job. The error is not present on the other node where the same job is being executed, however the issue with the missing tasks also exists there regardless.


I do not know if there is a global task log for the job (or where to find it), I have only been using Proxmox for a few weeks, so I'm still new to the whole thing.
I run a cluster of three nodes with two different backup jobs, spanning all three nodes.

A little extra information

On PVE I have mounted two different datastores, which basically are pointing to the same datastore on PBS, but with different namespaces, in order to differentiate retention options between the two jobs. Each job points to its own datastore. The daily one has no issues with Task logging.

PVE:


PBS:

 

[dcsapak]

I do not know if there is a global task log for the job (or where to find it), I have only been using Proxmox for a few weeks, so I'm still new to the whole thing.

what i meant is in the bottom task log list in pve, there should be a 'backup job' task that contains all the individual backup tasks too

btw. if you have multiple nodes, the task log will only exist on the node where the action happened, so when the vm is migrated to a different node, it won't migrate the task logs there...

 

[Grus]

what i meant is in the bottom task log list in pve, there should be a 'backup job' task that contains all the individual backup tasks too

btw. if you have multiple nodes, the task log will only exist on the node where the action happened, so when the vm is migrated to a different node, it won't migrate the task logs there...

Bingo. There they are. Both of the nodes with vms included in the missing job has a single “Backup Job” entry in the global task list, which seems to contain the complete sequential backup log of the included vms. It still puzzles me whey they don’t show up under the individual vms?

An interesting observation here is, that job 1 which contains two vms - each placed on a different node, thus only one vm on each node is affected - gets the task log to the individual vm. But for job 2, where multiple vms on each node are affected, the log never gets to the vm. But only the global log.

Also notice the difference between the backups run at 01.00 (single vm) and at 02.00 (multiple vms). The first lists the vm id, the other just announces that a backup job has run.

 

[dcsapak]

that's probably just an optimization that a 'backup job' (which normally has multiple vms to backup) with only vm gets started as a 'single vm backup' either way, we only start one task per job, and not one global task + a task per vm

 

[Grus]

that's probably just an optimization that a 'backup job' (which normally has multiple vms to backup) with only vm gets started as a 'single vm backup' either way, we only start one task per job, and not one global task + a task per vm

That makes sense. With my above observation in mind, I assume that the reason why the task log never gets to the VM, in case of multiple VMs in the backup job, is that the log and job in this case, is attached to the PVE node and not the VM. Or is it a bug?

 

[dcsapak]

in case of multiple VMs in the backup job, is that the log and job in this case, is attached to the PVE node and not the VM. Or is it a bug?

no the difference is that the 'multi backup job' does not has the vmid in it's id so it's not attached to the vm (without opening every task log it would be impossible to map that correctly)

 

[Grus]

no the difference is that the 'multi backup job' does not has the vmid in it's id so it's not attached to the vm (without opening every task log it would be impossible to map that correctly)

That makes sense. Thanks for the clarification.