LifePreserver replication issues #1434
Comments
|
Has anyone tried LP replication successfully recently? |
|
Im using it successfully on a couple of my systems right now.
Sent from TypeApp
…On Jun 6, 2017, 5:03 AM, at 5:03 AM, danboid ***@***.***> wrote:
Has anyone tried LP replication successfully recently?
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
#1434 (comment)
|
|
Ah good - so I should be able to get this working if it works for you. Do you see any mistakes in my config? The handbook only covers replicating to a FreeNAS box but I should be able to replicate to my ZFS-based FreeBSD 11 box, right? Do you plan to add the notifications/warnings and error messages I suggested to LP / sysadm? Should I see a progress bar or some form of status when it is replicating? Thanks |
|
I should also mention that I do have a /var/log/lpreserver dir but it doesn't contain anything, even immediately after trying to replicate or initialize. Does LP default to using SSH keys if you don't input a password when you add your target dataset? |
|
same issue here except, the other box is a freenas |
|
Thanks to members of the discourse board, I think I understand LP a bit better now. There seems to be a few issues that need fixing here. One thing I was unaware of, was the need to copy the private SSH key from the root users .ssh dir on the TrueOS box into the .ssh dir of the user you are backing up to on the server. Yes, LP depends on SSH keys and doesn't seem to offer any support for using SSH passwords. Seems really obvious now. The original bug/missing feature here is that I was pushing 'Start' under the Replication tab of the LP GUI client, thinking this was how you would perform a first replication. Replication then snapshots, I expected. Turns out you have to create at least one snapshot before you can use replication so the user should be told this if they try to use replication before any snapshots have been created. Looking at Fig 11.6.2 on: https://sysadm.us/handbook/client/sysadmclient.html#life-preserver It seems two main things are missing before we can use LP SSH restore: 1 - The ability to load a SSH key 2 - The ability to enter a dataset to restore from. I want to be able to backup multiple TrueOS installs to individual datasets on one ZFS based FreeBSD (or FreeNAS, or Linux ZFS) server. At least the first was present and working in the PCBSD installer, I'm told although I never tested it. |
|
maybe @kmoore134 can jump in again here? |
|
It would be cool if there was an LP equivalent to ssh-copy-id that would create a dataset, set the zfs permissions on it and copy the key - everything except create a new user. |
|
As I said in my recent BSD Now interview, I think LifePreserver (in particular easy network restoration of full installs) could be the other killer feature of TrueOS, alongside its great support for boot environments such as its unique boot env install feature |
|
Just having a look at this again now, the docs don't seem to mention using SSH passwords as being an option, yet when setting up replication targets we have a password entry box? As previously mentioned, the SSH restore wizard screen of the installer lacks any way to enter a password. I'm not going to post again until kmoore134 has his say. |
|
Still no reply from Kris! There's another suggestion I want to make a note of here before I forget. I have made requests across multiple github issues on here for making improvements to the TrueOS installer so as to make it possible to dual/multi boot TrueOS on a single disk, BIOS-based machine without resorting to workarounds like I have. In addition to these, I would also like to see the LP restore wizard give users the option of either restoring your TrueOS partitions to the exact same sizes as they were originally or to expand the TrueOS partition(s) to use all available space on the drive. This would require that partition sizes are noted when snapshots are created, of course, and would be a necessary feature for anyone wanting to multiboot on a single drive and use the restore wizard. I am presuming it doesn't do this already? |
|
A discourse user (twschulz) has submitted a pull request that he believes should fix SSH LP restores: |
|
Those PRs have been pulled in - just need to test with the next UNSTABLE build. |
|
I am unsuccessfully testing with the TrueOS-Desktop-2017-10-16-UNSTABLE-x64-DVD.iso |
|
I've been promising to get to these lp issues for a while now. I'll code up a ssh-key exchange option to make this 'easy' |
I run FreeBSD 11 on my home NAS so I want to use that to backup my TrueOS install via Lifepreserver but its not working for me.
First, I checked I could login as my normal user via SSH with a password. I had disabled SSH access with passwords on my FBSD box but I thought there was more chance of LP working if I just used passwords.
I created a new dataset on my FreeBSD server like so:
zfs create -o compress=lz4 data/sys/true76then I set the ZFS permissions as covered in the TrueOS handbook:
zfs allow -u dan atime,canmount,clone,compression,create,destroy,hold,mount,mountpoint,promote,receive,rename,send,userprop data/sys/true76I then opened LP on my TrueOS laptop and created a new replication target. It's at this point I'll highlight I'm not using a standard SSH port so that was entered along with the other details of my FreeBSD server. Then, when I click 'Start' - nothing happens. No errors, no warnings, no progress bars... nothing. Exactly the same happens if I click 'Initialize' - nothing at all.
If a connection cannot be establish when clicking 'Start', I expect a notification of some sort to alert me of the error, otherwise I would hope to see a progress bar or something somewhere to let me know a replication is underway. Clicking 'Initialize' should also print an error if the connection cannot be established and if it can, I would expect it to ask for my confirmation before it wipes a dataset / replication target.
The text was updated successfully, but these errors were encountered: