Effects of editing /etc/pve/storage.cfg monhosts on live datastores

[smetw]

I'm running a PVE 8.4 cluster. The datastore our VMs use is an external Ceph cluster running 19.2.2.

Yesterday, I found a "design flaw" in our setup. In /etc/pve/storage.cfg, there's this:

rbd: pve
    content images
    krbd 0
    monhost mon.ourdomain.com
    pool pve
    username blah

mon.ourdomain.com resolves to all the IP addresses of the Ceph nodes that run mon daemons. Now, the problem is that I now also want to migrate our DNS servers to run on Proxmox, and they will also use the RBD pool called "pve". So, what happens in case power goes down and we have to boot our environment from scratch? Right, the Datastore holding the RBD images of every single VM we run well never get mounted. So not a single VM (including our DNS servers) will be able to boot because there's no DNS.

To mitigate, I would like to change 'mon.ourdomain.com' to a space separated list of the 5 IP addresses our mons run on in /etc/pve/storage.cfg.

Question:
Can anyone with technical proficiency tell me what will happen if I edit /etc/pve/storage.cfg to reflect the changes I want to make? Can I be absolutely sure it will not affect the RBD pool already "mounted"? I want to exclude that whatever happens, it's virtually impossible to crash all running VMs on that "pve" RBD pool by (correctly) manipulating the /etc/pve/storage.cfg file.

Secondary question: I'm not sure if I can give the monhost directive both mon.ourdomain.com AND a list of 5 IP addresses.

 

[Onslow]

I don't know the answer to your question, but I think a workaround could be placing multiple lines for mon.ourdomain.com in the /etc/hosts

 

[smetw]

I've been thinking about that too. But on the other hand, I'd like to keep config files as vanilla as possible. You never know if you mess up some internal scripting by putting records that are not supposed to be there. (unlikely but still).

Meanwhile, I'm keeping it as is. If we get hit by a power outage, I know I have to intervene to get our environment to boot and at the same time rectify the issue.