New System Images and backup restore hell - no images detected, no backup restored

Hello. I will post these as a public issue, because I hope someone encountered them already. They are dealing with the same thing, migration, a must for virtualization.

So I am trying to migrate from an "old" Cloudmin Pro setup to a new one. I got two servers, both are making use of KVM via their own centralized NFS storage. Why? Because the working server and storage are old, and I intend to use them for failover.

So I installed Cloudmin Pro on the new server, hook up the new storage, configured both; all is fine, I can see my exports, mount, read/write in any directory. And I have a similar setup working.

  1. New System Images... not showing my images? So I figured I could make some images on the old server (and storage) and just copy those on the new one, in the coresponding directory. And install new VMs from there. Problem is they do not appear in Cloudmins interface, nor when creating a new KVM VM, or simply in the Cloudmin settings > New System Images list. I downloaded just for testing a Centos 7 image. Well, well: it does appear in my same /mnt/3-IMGs directory besides my carefully built and useless images. And in Cloudmins interface I can see this (downloaded) one but not mine... I couldn't find a "scan" command in the docs if Cloudmin is not looking directly for the files. Ergo, the directory is properly set, the server reads and writes files, but it is a Cloudmin thing. What can I do?

  2. Restoring backups from the old Cloudmin (same, lates version to date) is not working also. I mean they do somehow - the interface seems to be working, and I can see a dd process that never stops. How can this be? I have VM of 150GB and (tried this yesterday with another VM, the same thing) when I look at the .img file it just grows... I mean right now it's 513GB???

So now I have no option to move my VMs, because you just can't copy them to another system (still, why? is this still an issue?) and import them... :( Please don't tell me to start things all over again...

What use has the backup option then? How can one restore stuff? For example If my working setup will burn in flames (started from scratch the VMs if wondering)?

PS: the hardware is fine, new, storage wise I use ZFS, enterprise grade SSDs, the network is 40Gb/s QSFP Twinax between each server and it's storage and 10Gb/s SFP+ for the internal network between the servers.

Status: 
Active

Comments

fakemoth's picture
Submitted by fakemoth on Wed, 07/18/2018 - 05:41 Pro Licensee

dd if=/dev/zero of=/mnt/1-VMs//w1.somedomain.ro.img bs=32768 is the command and if I kill it, the w1.somedomain.ro.img starts again form 0, now 27GB and it spawns:

dd of=/mnt/1-VMs//w1.somedomain.ro.img bs=32768

and a bit later

gzip -d -c /mnt/4backs/Cloudmin/h1.somedomain.ro/18-07-2018-09-44//w1.somedomain.ro.gz

???

fakemoth's picture
Submitted by fakemoth on Thu, 07/19/2018 - 01:33 Pro Licensee

It eventually finishes AFTER I kill the first dd command but I see only two files w1.somedomain.ro-eth0.sh and w1.somedomain.ro.img and of course it throws the error, notice the network one, the other I am used to it:

Updating network interface bridges ..
.. failed to list network interfaces : Failed to mount filesystem : None of the partions in the disk image /mnt/1-VMs//w1.somedomain.ro.img could be mounted and contained the /etc directory. This may be due to the use of RAID or LVM.

Also in the systems list it shows 536.63GB of space for w1.somedomain.ro (probably till I killed the process) but in the Resources > Manage Disks it prints corectly, 150GB...

fakemoth's picture
Submitted by fakemoth on Wed, 07/18/2018 - 12:41 Pro Licensee

  1. Solved, didn't test if it starts yet - I will do it tonight: somehow I never used Import System Image, this is the stuff, right?
  2. Something is very wrong with restoring backups, just had to kill another one after about 6 hours of doing... I don't know, some huge file? On SSDs?

I understand why it needs to dd me to death - because some prefer LVM.

I will never understand though why Cloudmin doesn't treat us, the "file people" in a different way - just let me copy my file/directory if you need some other files and start the VM... Don't check for grub, volumes, mounts and the kind at every creation, imaging, backup, restore and image import. Don't compress and recompress again, If I don't feel like wasting time - took 4hours to get a VM, after 2 days of testing, scp copying aside... that is not "virtualization agility" :)

Just (keep on) saying :)

So to address your first point - the simplest solution may be just to copy the image file(s) from /var/webmin/server-manager from the old to the new Cloudmin systems. This would then allow you to create a new VM from those images, and everything should just work.

fakemoth's picture
Submitted by fakemoth on Fri, 07/20/2018 - 15:14 Pro Licensee

Thank you I will try this too. What about the backup issue, where should I look? It really scares me... BTW no memorable errors in miniserv logs.

Back to the first point: the Cloudmin status of the VM it's in a state of flux: No SSH, ping failed and stuff. Now I can ping from the "internal" network between the servers, but not from outside. Nor can the VM ping anything outside, really weird, stopped the firewalls, checked the IPs, the routes... Should I start another issue, a private one?

Have you tried restoring a backup from the command line with the cloudmin restore-backup command on the master? If you are restoring a huge backup from the UI, it may be timing out in the browser.

fakemoth's picture
Submitted by fakemoth on Sat, 07/21/2018 - 01:47 Pro Licensee

I will - but the job should be running, though if that's the case. Managed to make another VM from an image as I said. I am opening another private issue

Hi, I have the issue restoring a virtual machine on 10Gb with the "New system named" option I have killed after dd process had create a 300Gb file, but i have not resolved the issue.

THanks Alberto I use last version of cloudmin on Devuan ascii HP Microserver

sennalb - can you open a separate ticket for this? Yours sounds like a different issue..

Hi, I get the exact same problem as mentioned by Alberto and for which he opened an issue.