Proxmox installer question
- Thread starter azop
- Start date
Thank you, one last question...would you recommend hardware raid with 2 x 2TB drives with no zfs caching or 2 x 2TB drives with a 160GB SSD for zfs cache? I've not used zfs before but I can follow instructions.
There would be around 8 VM's (1 db server with moderate i/o, 6 general servers with low i/o, 1 server with moderate i/o and one windows VM). The server would be a Xeon 1230v3 w/ 16GB of ram
Thanks
There would be around 8 VM's (1 db server with moderate i/o, 6 general servers with low i/o, 1 server with moderate i/o and one windows VM). The server would be a Xeon 1230v3 w/ 16GB of ram
Thanks
zfs needs a lot of ram for running great. so with 16 GB, you are a bit limited here.
if you do not use or need zfs advanced features (compared to hardware raid), use a hardware raid controller and you have 16 GB ram for your VM (and host).
if you do not use or need zfs advanced features (compared to hardware raid), use a hardware raid controller and you have 16 GB ram for your VM (and host).
Thanks, I can add another 16GB of memory to the server. I like the features of zfs so I'd prefer to use it.
Would it be unreasonable to partition the 160GB drive and use some of it zil / l2arch and the other for a VM?
What would you recommend for a zil / l2arch sizes with 32gb of ddr4 memory?
Sorry for the questions, I appreciate the pre-sales support.
Would it be unreasonable to partition the 160GB drive and use some of it zil / l2arch and the other for a VM?
What would you recommend for a zil / l2arch sizes with 32gb of ddr4 memory?
Sorry for the questions, I appreciate the pre-sales support.
Consider, if I'm not wrong, that unfortunately the Proxmox installer creates ZFS root file system setup so you have VM image file on it, so forget snapsthots at zfs level (you have to use qcow2 format for that, and then you will have a COW fs upon a COW filesystem...), so some ZFS features are not available in any case (i.e. snapshot and use zfssend for instance)
Don't know if there is a different way to take full advantage of ZFS starting from default Proxmox installer setup
Now I'm confused, so if i use 2 x 2TB drives + ZFS for the root file system I can't use snapshots?
What happens if one of the 2TB drives dies? Will things 'just work' or do I have to boot into rescue and fix things manually?
Sorry, but you are totally wrong... Such ZFS setup is fully functional, so I do not really know what you talk about here.
I mean, with a ZFS bare metal installation if I create a VM with RAW disk I can't have snapshot, I need to create as qcow2 file type, so then the snapshot is done through qcow2 mechanisms and not zfs (I guess). If instead I install proxmox as ext4 and add storage as ZFS type, when I create a VM AFAIR zfs volumes are created, and those are snapshotted through ZFS features, and I can use zfs commands to create snapshot.
Here I've 3 running vm with qcow2 disks (zfs bare metal)
Code:
root@prox02:~# zfs list
NAME USED AVAIL REFER MOUNTPOINT
rpool 424G 475G 96K /rpool
rpool/ROOT 417G 475G 96K /rpool/ROOT
rpool/ROOT/pve-1 417G 475G 417G /
rpool/swap 7.44G 482G 308M -
root@prox02:~#here instead a test I did with added ZFS pool and a vm created on it
Code:
# zfs list
NAME USED AVAIL REFER MOUNTPOINT
raid1poolzfs 5.01G 26.2G 30K /raid1poolzfs
raid1poolzfs/vm-100-disk-1 5.01G 26.2G 5.01G - <------------------------------
rpool 5.37G 25.9G 144K /rpool
rpool/ROOT 1.25G 25.9G 144K /rpool/ROOT
rpool/ROOT/pve-1 1.25G 25.9G 1.25G /
rpool/swap 4.12G 30.0G 228K -If you create qcow2 VMs you have snapshot, AFAIU is provided by qcow2 and not ZFS, that is what I meant about "not all zfs features available"
If one of your HD dies, is not related to snapshot, if you configured it as RAID1 it will survive in degraded mode
or just use the GUI. no missing feature here.
http://pve.proxmox.com/wiki/ZFS#Using_ZFS_Storage_Plugin_via_Proxmox_VE_GUI
Even more confused! There is already a pool, "rpool", how can I create one more? Something like "zpool create tank" needs to specify one or more disks, should I specify the same disks already used by "rpool"? Could you please provide a sample, because telling a newbie general stuff without a concrete example is almost useless
Thanks a lotNow I've got it and thanks, works (this way if I select that storage a ZVOL is created for each disk, great!). But now I've 3 more questions: a) why should I create a new file system instead of using one of the already available? (and in doing so, which one is better among "rpool", "rpool/ROOT" or "rpool/ROOT/pve-1"?) b) I find the GUI misleading with the field "ZFS poool" while a "ZFS File System" is required instead. 3) why not having the installer set it up by default? I think using the FS and not ZVOL has only disadvantages, or am I wrong? (in any case I will edit the wiki about this topic as soon as I will have some free time).
Thanks once again for the support and the great work!
One problem with the out-of-the-box ZFS setup of Proxmox, and this has not been changed to my knowledge. It puts swap on a ZFS volume, and this is fairly certain to crash the host after a little swap usage in Proxmox. This also puts a little doubt on the behaviour of ZVOLs generally under memory pressure. I detailed this in:
https://forum.proxmox.com/threads/zfs-swap-crashes-system.25208/#post-126215
https://forum.proxmox.com/threads/zfs-swap-crashes-system.25208/#post-126215