LVM – thin-pool only 43% of data but almost all PE used

So I’ve got a bit of a situation, I’ll try to keep it brief!
Initial server setup:

2x drives with soft RAID1
LVM thin-pool setup on the md3 RAID device

Then the md3 (raid) device gets full so in my haste to fix the problem (knowing I can’t get an additional drive added at the DC) I removed the 2nd disk from RAID and setup a new PV on it. The remaining disk was left in RAID (just 1 drive) as a PV on the VG. I then extended the thin-pool to this “new” PV to effectively double the storage within the thin-pool for use by the VMs.
The mistake I made when doing this is (probably one of many), I forgot to completely format the drive so the previous LV structure was still present when I extended the thin-pool but I didn’t realise this until some time afterwards. However from my research I understand that when you set up a new PV on an existing volume, it’s a destructive act so despite the previous LV structure still showing the new drive is being used by LVM and it is spanning any new data across the two PV’s.
Not a perfect solution but it resolved the initial issue of no space being available.

Now I’m left trying to resolve this issue:
The thin-pool across the two drives is only 43% full of data + meta data

[root@XXX ~]# lvs
  LV                                        VG  Attr       LSize   Pool      Origin Data%  Meta%  Move Log Cpy%Sync Convert
  test                                      vg1 Vwi-a-tz--   5.00g thin_pool        20.00
  thin_pool                                 vg1 twi-aotz--   3.14t                  42.63  28.69


[root@XXX ~]# pvdisplay -C
  PV             VG  Fmt  Attr PSize  PFree
  /dev/md3       vg1 lvm2 a--  <1.58t     0
  /dev/nvme0n1p4 vg1 lvm2 a--  <1.58t 15.73g
[root@XXX ~]# vgdisplay -C
  VG  #PV #LV #SN Attr   VSize  VFree
  vg1   2  59   0 wz--n- <3.16t 15.73g

It would appear that even though there should be enough space left across the two drives, the PE’s that were “left over” (when I forgot to format) are still being treated as if they’re in use so I barely have any free PE’s left.

When you do a lvs the LVs are not duplicated in either quantity or LV size, they’re exactly as I would expect them to be and the usage of the LV’s again is what I would expect them to be, not duplicated or doubled.

Any ideas on how to resolve this highly frustrating and self-made situation?

Any help would be enormously appreciated.

Go to Source
Author: papa_face

Server is down after changing machine type in Google Cloud Platform

I changed my machine type(GCP) by stopping the instance, after increasing the machine type I restarted the instance and now server stopped working, the website is down now.
Upgraded from 1 vCPU, 1.7 GB memory to 2 vCPU, 4GB memory.
And if I select other than 1 vCPU, 1.7 GB memory, the webserver doesn’t respond.
Also, I can’t access SSH from a web browser it gives me an error that “Connection Failed. We are unable to connect to the VM on port 22.”


Go to Source
Author: Sandeep Kumar Giri

How do I manage xen-hypervisor using virt-manager client?

I’ve installed xen-hypervisor 4.11 on my debian 10 desktop. But, when I add xen connection to my virt-manager it shows this error.

Unable to connect to libvirt xen:///.

Failed to connect socket to '/var/run/libvirt/libvirt-sock': Connection refused

Verify that:
 - A Xen host kernel was booted
 - The Xen service has been started

Libvirt URI is: xen:///

Traceback (most recent call last):
  File "/usr/share/virt-manager/virtManager/", line 1012, in _do_open, self)
  File "/usr/share/virt-manager/virtinst/", line 138, in open
  File "/usr/lib/python3/dist-packages/", line 104, in openAuth
    if ret is None:raise libvirtError('virConnectOpenAuth() failed')
libvirt.libvirtError: Failed to connect socket to '/var/run/libvirt/libvirt-sock': Connection refused

Do I need a IOMMU compatible motherboard to use Xen-hypervisor? or something else?

Go to Source
Author: Michael Mosot