Problems with my VMs

[javiermi]

I need desperately  some help

Today, while I was creating a new VM, suddendly my VMS decided to pause and was not able to resume them

It seems the cache disk is full, so tried to run this, as per some other forums:

btrfs balance start -dusage=75 /mnt/cache

When this failed, I decided to reboot, but started and all my VMS had dissapeared. I tried to recreate them from the xml files, as the files seem to be there, but it failed with a message about read-only filesystem

I have tried to stop/start vm manager, but now it fails to start saying:  "Libvirt Service failed to start"

AS I said I need at least one of the VMS to be working - I do have some configuration backups - as well as the original images

 

At this point I don't know what do do!

I appreciate any help you can provide

 

Find enclosed diagnostic file

 

Javier

 

unraid-diagnostics-20200525-2226.zip

Edited May 25, 2020 by javiermi
2020-05-25 21:13:18.953+0000: 10211: info : libvirt version: 5.10.0 2020-05-25 21:13:18.953+0000: 10211: info : hostname: UnRaid 2020-05-25 21:13:18.953+0000: 10211: error : catchXMLError:767 : /etc/libvirt/qemu/ELK.xml:1: Document is empty (null) ^ 2020

[JorgeB]

9 hours ago, javiermi said:

btrfs balance start -dusage=75 /mnt/cache

This is only for when the cache if fully allocated but there's still free space, your pool is completely full so you need to move or delete some files.

 

If the VMs are Windows this helps to keep the vdisks as small as possible.

[javiermi]

Thanks for your reply

I have managed to recover the VMS, but only because I had a backup of the xml files, so I managed to rebuild them 

You are right: the cache drive was full.( BTW, the issue was with Linux VMs, not Windows). Only now I have started looking at how the mover works - and have set a daily schedule to move stuff out during the night

I did a manual move after changing the appdata and domains shares to "use cache yes", and this cleared about 30Gb our of 120Gb, so that was great

However, I have created a new Linux VM, and now the cache disk started to fill up again

Now I know I have to keep an eye on the cache size, but I think from the documentation is not very clear on how the cache drive works, and to be honest I think this is a major "flaw" in how it has been implemented: what's the point to have a HA Server ( unraid) that fails because  a disk is full?

I'm thinking of buying an extra SSD to expand the cache disk, but I have two questions:

 - If I want to expand it (not replace) , do I need to buy a drive of the same size? or can I buy a bigger size drive

- Once I expand the disk, will Unraid try to fill up the disk space to the max ( again)?

 

Thanks

[JorgeB]

You can enable system notifications and will receive a warning if the cache gets to the set warning thresholds.

 

12 minutes ago, javiermi said:

If I want to expand it (not replace) , do I need to buy a drive of the same size? or can I buy a bigger size drive

You can expand with any size, using the single profile, see the FAQ.

 

12 minutes ago, javiermi said:

Once I expand the disk, will Unraid try to fill up the disk space to the max ( again)?

It will fill it up to the size of the vdisk, but the vdisk itself can use more space than expected with time if not using unmap, though only have experience configuring that for Windows VMs.

[javiermi]

Johnie 

Thanks for your quick reply

regarding the cache expansion question, I know I can set a raid 0  ( stripe set), but my understanding is that by default it uses raid 1 ( mirror), then you need to change it. Is this the case?

 

Thanks

[JorgeB]

19 minutes ago, johnnie.black said:

see the FAQ.

https://forums.unraid.net/topic/46802-faq-for-unraid-v6/?do=findComment&comment=480421