Proxmox datastore not cleaning space

[ady106034]

Hi,

Below is the scenerio.
1. Have proxmox node 7.3.3.
2. Created VM of promox backup server.
3. Have thin pool in main node.
4. Given a 8T out of 32T to promox backup server mounted as /datastore. This datastore is added proxmox backup server to take vms backup of our vms.
5. Now space is getting filled Up so I did clean older backup more than 1 T. But still the database store showing filled. Executed the fstrim as well garbase collection.
Fstrim showing Output like 800G is trimmed but still its showing usage as earlier.

Discard is enabled.
/dev/vdb xfs 7.9T 7.1T 826G 90% /datastore

 

[ady106034]

Can anyone help me?

 

[Neobin]

See:
https://forum.proxmox.com/threads/do-a-real-cleanup-of-unused-files.125220/post-546028

Can anyone help me?

This is a (official) community forum. You can not (necessarily) expect an answer within only 44 minutes...
If you need guaranteed support in a timely manner, there are support subscriptions available with e.g. the "premium" one including the fastest response time of 2 hours within business hours:
https://proxmox.com/en/proxmox-backup-server/pricing

 

[ady106034]

Its now more than 24 Hours. Can anyone help?

 

[proxstormer]

I have same pattern on all my pbs, the datastore will always end filling up, no matter how many backups you delete, the GC doesn't remove all old chunks, I allways end up deleting datastore, and starting again.

 

[Dunuin]

Have proxmox node 7.3.3.

You don't like fixing vulnerabilities?

I have same pattern on all my pbs, the datastore will always end filling up, no matter how many backups you delete, the GC doesn't remove all old chunks, I allways end up deleting datastore, and starting again.

Did you wait 24 hours + 5 minutes between running the prune and GC? Otherwise GC won't free up pruned backups.

 

[proxstormer]

Yes I wait 24h5m, I will make a test setup this WE to show the behavior.
I see the GC report with the Number of GB removed and the Number of GB pending to be removed.
The case is in each iteration the space is always consuming, is like a memory leak but on the disk space.

 

[Dunuin]

So number of pending chunks will go up each day or does it remove the chunks but the storage won't free up the space of the deleted chunks?

 

[proxstormer]

It removes the chunks and storage removes space, till here all is right.
The problem is that space that is used in the storage is more that the used in chunks, having that the datastorage is on a drive alone and I having only one datastorage.
I saw this problem on all pbs I installed, the space used in chunks reported on the GC differs the space used on datastorage.
Can you check on your pbs?