Base Copy / VDI not merged
-
@aoprea said in Base Copy / VDI not merged:
Jan 14 09:19:25 xen-scada SMGC: [5206] Got other-config for 7995044dVHD: {'leaf-coalesce': 'offline'}
Jan 14 09:19:25 xen-scada SMGC: [5206] No space to leaf-coalesce 7995044dVHD (free space: 97983135744)
Jan 14 09:19:25 xen-scada SMGC: [5206] ...but enough space if skip snap-coalesceYes, that's the issue. It says you have 91GiB free but you need 200GiB free to make the coalesce.
-
@olivierlambert
so it's not enought to remove seccond VM (10Gb) to finish ?what do you recomend to do ?
-
shut down big VM, coalesce will start and will be finish ? how can i estimate what time ill be off ?
in this time of poweroff big vm, can i try to export'it VM ? or i have to wait to finish coalesce if its start ?
if i shut down big vm, will be start again if coalesce not start or not finish ? -
add a seccond hdd (1 TB) to phisical machine host and expand Local Storage SR.
This can be made without reinstall xcp-ng ?
or is better to reinstall xcp-ng with these two hdd's attached ?
ps: i have a last good backup of big vm before this coalesce problem
sorry for this many questions
thank you for your help -
-
I would switch to local ext instead of LVM so you can rid of this problem forever.
- Shutdown the VM, backup it to get the last version.
- Remove the local LVM SR
- Recreate the local SR as ext
- Restore the VM
- Enjoy
-
Because it is important that the downtime to be as short as possible,
I'm thinking of first, to preparing a second phisical host machine with last good backup of vm and put'it online.
In this period, on principal machine i will go with steps described, or reinstal from the beginning ussing SR as local ext with 2 hdd of 1 Tb each.the problem what i have is in the first step:
1. Shutdown the VM, backup it to get the last version.can be backup'it after shutdown vm ? or after coalesce will be start and finish and then backup'it ?
and go to rest of stepsthank you and sorry for bothering you with questions
-
It's very likely your SR is so full than any snapshot won't do it, that's why you need to backup while it's off to avoid any snapshot. Sadly, when you are using a thick pro SR with that much single large disks (vs the total size on the SR), you can end in this situation.
Now, reducing the downtime can be achieved many way, but keep in mind there's probably not enough space for snapshots, so no export/backup while the VM is running.
-
ok thank you
my question mark is why everything was fine for a long time period (backup's they were done regularly every 3 days),
and suddenly it stopped working to do backups ? -
It worked as long as you had no coalesce issues or lag, or a snapshot wasn't removed or garbage collected correctly. In short, you had very little room for error. Using ext will give you all that space you need
-
For to switch to Ext SR what wil be a good aproach of this:
-
separatly hdd for instalation of xcp.ng os (let say 250Gb ssd) and separate one or two hdd for local Ext SR (2 x 1Tb hdd)
-
two hdd of 1Tb but install of xcp.ng OS will be in one of hdd partition (250Gb) and a rest 750Gb partition + 1Tb second drive as Ext SR local
Like i show my vm hdd size is 400Gb (aprox 300Gb phisicaly ocupied) but.in long-long future slowly increase to 450-500 Gb maxim hdd size.
Thank you
-
-
It's a matter of personal preference, but I admin I like to split the system disk (small SSD) with the SR (using whatever technology but bigger capacity). Because if the SR dies, your system is still up, and if the system drive dies, your data is still here. A bit more flexible.
-
@olivierlambert said in Base Copy / VDI not merged:
It worked as long as you had no coalesce issues or lag, or a snapshot wasn't removed or garbage collected correctly. In short, you had very little room for error. Using ext will give you all that space you need
what I don't understand is: when use Ext SR in same condition (sr 890gb vm 400gb) when do a backup it will no longer be created base_copy ?
this base copy is created only when use lvm SR ?Or what is the difference between Ext/lvm regarding backup process
thank you
-
Ext is thin pro, meaning yes, you still have the base copy, but the active disk will be only few KiB instead of the full size in LVM. It's not backup related but storage thin vs thick.
-
Ok i understand now
What do you think about this aproach of machine configuration:
- 1 hdd 200Gb as xcp.ng os
- 1 hdd 1Tb as primary Ext SR Local Storage with 2 vm running (principal vm and xoa vm)
- 1 hdd 1Tb as seccondary Ext SR Local Storage where i will do Continuous Replication of that vm
Of course i will do and a full backup of vm to a smb share, and from here a new backup to external hdd
I know that CR will be more efficient if it was done to another machine in another location ... but in the next future will be.
Thankyou
-
I was about to tell you a warning about doing on CR on the same host but you are aware about the risk doing that, so I have nothing to add, it looks OK
-
yep i know but i think it is much better to have 2 Ext SR each with 1 hdd than to have 1 SR with a raid with 2 hdd, regarding downtime in case of failure disk
