2nd local disk available space is zero

[jagan]

Hi,

In one of my 4-node cluster, the node is having 2nd local disk (4TB directory type). It is showing available space as zero. But when I check in my 2 windows VMs the both the VMs still having 1.5TB of free space.

Please help me how can fix this issue.
Thanks in advance.

 

[dcsapak]

where does it show available of zero? can you paste the output of df ? also the vm configs and storage configs would be interesting

 

[jagan]

where does it show available of zero? can you paste the output of df ? also the vm configs and storage configs would be interesting

Please find the attached screenshots df, vm config (VM1, 2 ,3), storage config(/etc/pve/storage.conf) and data size in each vm

 

[dcsapak]

you have overprovisioned your storage
you have set 9728+900+900G = 11528G
while your storage only has 11060989840k = 10549G

your vms will not be able to write anymore and probably produce damaged disks

if you overprovision storage you should never let it run full!

 

[jagan]

you have overprovisioned your storage
you have set 9728+900+900G = 11528G
while your storage only has 11060989840k = 10549G

your vms will not be able to write anymore and probably produce damaged disks

if you overprovision storage you should never let it run full!

Since it support thin provision I have over provisioned to 3 windows servers. But I have not allocated all the space in windows disk manager. And VM1 is having 2.5TB free space , VM2 is has 800GB free space & VM3 has 180GB free space. In total we have more than 3TB of free space in 3 VMs. We don't understand that for "df" command it is showing as 100% utilized, free is 0%.

Can you please help me to fix this error.

 

[dcsapak]

And VM1 is having 2.5TB free space , VM2 is has 800GB free space & VM3 has 180GB free space. In total we have more than 3TB of free space in 3 VMs.

this is irrelevant, as the data on the disk is not 1:1 what the vm sees, e.g. you write a big file in the vm and the delete it again, the space on the disk below may not decrease

you can set 'discard' on a disk and then call e.g. fstrim in the guest, this should minimize that issue, but in general => if you overcommit resources, monitor them, else you will encouter problems sooner or later

 

[jagan]

this is irrelevant, as the data on the disk is not 1:1 what the vm sees, e.g. you write a big file in the vm and the delete it again, the space on the disk below may not decrease

you can set 'discard' on a disk and then call e.g. fstrim in the guest, this should minimize that issue, but in general => if you overcommit resources, monitor them, else you will encouter problems sooner or later

I will check the same in my test environment and come back with results. Thanks for your reply.