This is my first round of testing LXC in PVE 4. Had to stop kind of early on... - Installed PVE 4 on ZFS RAID10 (4 disks) - Updated to latest packages + reboot - Downloaded Debian 8 LXC template - Trying to create CT on local storage Code: Formatting '/var/lib/vz/images/100/vm-100-disk-2.raw', fmt=raw size=10737418240 Discarding device blocks: 4096/2621440 done Creating filesystem with 2621440 4k blocks and 655360 inodes Filesystem UUID: 13b2c657-77b9-43d6-b1a9-6f4ca55f47dd Superblock backups stored on blocks: 32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632 Allocating group tables: 0/80 done Writing inode tables: 0/80 done mke2fs 1.42.12 (29-Aug-2014) Creating journal (32768 blocks): done Multiple mount protection is enabled with update interval 5 seconds. Writing superblocks and filesystem accounting information: 0/80 [COLOR=#ff0000]Warning, had trouble writing out superblocks.TASK ERROR: command 'mkfs.ext4 -O mmp /var/lib/vz/images/100/vm-100-disk-2.raw' failed: exit code 144[/COLOR] The RAW disks get created under /var/lib/vz, but the container creation process stops with Error: unexpected status. Apart from the solving the above problem, probably a more graceful way of handling errors during the container creation process would benefit users.
Ok, so after some forum reading I found out that I need to create a new storage (type zfs) for containers, with it's root set to rpool (otherwise it won't work). This is far from being user-friendly, as there are no warnings during installation or container creation, and people expect simple functions (like Create CT) work out of the box. I suggest that the installer create a storage called "containers" automatically if installation was on ZFS, and the container creator should either select it by default, or at least there should be a warning or error (red outline) on the container storage if set to local.
Actually the real bug here is that local on ZFS is tagged as being able to host "Containers", which seems to be wrong.
Okay, so storing containers on ZFS local storage does not work. Hopefully restoring OpenVZ backups to LXC containers stored on newly created ZFS storage will.
I can attest to that. I've always used "dir" on ZFS for containers before 4.0. Now that I have them separate, each on its dataset is beautiful. Actually, that made me migrate back from SmartOS
I guess you mean storage plugin. So if I understand correctly, I have to restore the LXC containers with the "--storage" option to the newly created ZFS type storage if I want them to work. So let's say my ZFS type storage is called "zfs-containers", then I would restore my OpenVZ backups to LXC with the following command: Code: pct restore 100 vzdump-openvz-100-2015_10_12-11_22_33.tar --storage zfs-containers You might want to add this to the wiki at https://pve.proxmox.com/wiki/Convert_OpenVZ_to_LXC
And, how do you do that? -- create a new storage, type zfs, with root set to rpool. I've got a zfs mirror SSD set for boot, but another ZFS mirror that I want to put basically everything on. I get the error and can't make a container. Thanks in advance.
What is ZFS plugin for LSC container? I did a search and can't find anything useful. Thanks in advance.
PVE's LXC Storage plugin - it is what is used when you create a new storage of type ZFS in the web interface (and subsequently use it in pct/qm commands, LXC and VM config files, etc)
Oh, I see. I just considered that part of the system, rather than something that I needed to add as a plugin, like Firefox and Chrome plugins must be downloaded and plugged in separately.
I've got a zfs mirror SSD set for boot, but another ZFS HDD mirror that I want to put basically everything on. I get the error "failed: exit code 144" and can't make a container or a VM disk image, "failed: exit code 1". Thanks in advance. root@pve2:/# zpool list NAME SIZE ALLOC FREE EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT rpool 111G 1.11G 110G - 0% 0% 1.00x ONLINE - z3tred 2.72T 190M 2.72T - 0% 0% 1.00x ONLINE - root@pve2:/# zfs list NAME USED AVAIL REFER MOUNTPOINT rpool 15.9G 91.7G 96K /rpool rpool/ROOT 1.10G 91.7G 96K /rpool/ROOT rpool/ROOT/pve-1 1.10G 91.7G 1.10G / rpool/swap 14.7G 106G 64K - z3tred 190M 2.63T 104K /z3tred z3tred/backups 96K 2.63T 96K /z3tred/backups z3tred/containerstorage 96K 2.63T 96K /z3tred/containerstorage z3tred/containertemplates 189M 2.63T 189M /z3tred/containertemplates z3tred/diskimages 96K 2.63T 96K /z3tred/diskimages z3tred/iso 96K 2.63T 96K /z3tred/iso root@pve2:/# pvesm status local dir 1 97287552 1157248 96130304 1.69% sambadisk1 dir 1 2930081788 2505639992 424441796 86.01% zbackups dir 1 2827814784 0 2827814784 0.50% zcontainerstorage dir 1 2827814784 0 2827814784 0.50% zcontainertemplates dir 1 2828008576 193792 2827814784 0.51% zdiskimages dir 1 2827814784 0 2827814784 0.50% ziso dir 1 2827814784 0 2827814784 0.50%
You need to add your zfs pools as "zfspool", not as "dir" storage in PVE. PVE will then create subvolumes for snapshots, container and vm images, .. automatically.
Ok, Thanks. In DataCenter, Storage, Add, ZFS, the only pool shown is /rpool and it's subs. My other ZFS pool didn't appear. I tried this and got an error: root@pve2:~# pvesm add zfs -type zfspool -pool z3tred 400 not enough arguments root@pve2:~# pvesm add zfs z3tered -type zfspool --pool z3tred missing value for required option 'blocksize' Finally, I got to this command which seems to be working: root@pve2:~# pvesm add zfspool z3tred -pool z3tred -content images,rootdir -sparse
