Ceph cache tier and disk resizing

S

satiel

Member
Jan 13, 2016
11
0
21
39
Hello,
i'm currently running a ceph cluster (Hammer), last weekend I implemented a cache tier (writeback mode) of SSDs for better performance.

Everything seems fine except for disk resizing.
I have a Windows VM with a raw RBD disk, i powered off the VM, resized the disk, verified that both ceph rbd and proxmox see the new expanded capacity.

When i powered on the VM i noticed something strange, the guest doens't see the new free space.
I tried to refresh and rescan with all the tools that i know but with no results.

I think that the problem is that the rescan of the disk is performed against the cache instead of base pool.

After three days the problem remains.

What should i do?
 
did you resize the windows partition inside the vm?
 
satiel said:
Hello,
i'm currently running a ceph cluster (Hammer), last weekend I implemented a cache tier (writeback mode) of SSDs for better performance.

Everything seems fine except for disk resizing.
I have a Windows VM with a raw RBD disk, i powered off the VM, resized the disk, verified that both ceph rbd and proxmox see the new expanded capacity.

When i powered on the VM i noticed something strange, the guest doens't see the new free space.
I tried to refresh and rescan with all the tools that i know but with no results.

I think that the problem is that the rescan of the disk is performed against the cache instead of base pool.

After three days the problem remains.

What should i do?
Click to expand...
Hi,
work for me on an ec-pool too:
Code: 
# fdisk -l
Disk /dev/vdb: 10 GiB, 10737418240 bytes, 20971520 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes

### expand from 10 to 20GB ###

# dmesg | tail -2
[  134.086819] virtio_blk virtio2: new size: 41943040 512-byte logical blocks (21.4 GB/20.0 GiB)
[  134.086825] vdb: detected capacity change from 10737418240 to 21474836480

# fdisk -l
Disk /dev/vdb: 20 GiB, 21474836480 bytes, 41943040 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Udo
 
Hi,
the base pool is a replicated pool as well.
I created another pool with the same crush rule without cache tiering, resizing is working as expected.
Maybe this is happening because i added the cache tier to an existing pool ( i did it with all VM powered on) ?
 
You must log in or register to reply here.
Top Bottom