Op 10-02-2026 om 19:00 schreef Yossi Zamana:
Hello CloudStack users!
I’m looking for guidance on a CloudStack recovery scenario.
We have an Apache CloudStack environment with *5 KVM hosts* using *Ceph
(RBD) as primary storage*.
Due to an issue with the server hosting the CloudStack Management Server,
the management VM (including the CloudStack database) was deleted.
We do have a *valid database backup*, however it is *not fully up to date*:
-
The backup includes *only 4 out of the 5 KVM hosts*
-
A few VMs are *missing from the backup*
-
The *fifth host* was added after the backup was taken
Important note:
*All VMs (including those missing from the backup) have been running
continuously for several days* since the management server was lost.
All VM disks still exist on Ceph, and all hosts remain connected to the
same Ceph cluster.
So to get it correct: You have a certain backup schedule and before your
next backup run of the SQL database you made modifications and then it
crashed, correct?
Our questions are:
1.
Is it safe to restore the CloudStack database from this backup and then *add
the fifth host back* to the existing cluster?
2.
Well, no. As the UUID will be different. I would enter it into the
database manually to make sure it has the right UUID. This is a process
where you need to be cautious.
What is the recommended or supported approach to *recover or re-register
VMs that exist on Ceph and are running on the hosts but are missing from
the CloudStack database*, without impacting running workloads?
3.
Manual insert into the DB. Many tables where you need to insert data.
Are there any best practices for handling a *partial database restore*
in a Ceph-backed CloudStack environment?
Not particular for Ceph. You just need to manually add entries into the DB.
Our goal is to resync CloudStack with the existing infrastructure *without
data loss or service interruption*, as much as possible.
Thanks in advance for any advice or experience you can share.
Good luck!
Wido
Thanks
