On Monday, September 2, 2019 at 6:11:58 AM UTC+2, Andrew David Wong wrote:
>
> -----BEGIN PGP SIGNED MESSAGE----- 
> Hash: SHA512 
>
> On 01/09/2019 3.46 AM, 'Heinrich Ulbricht' via qubes-users wrote: 
> >> Personally, I would just stick with this. In other words, I would treat 
> >> the new Qubes installation as completely new and use qvm-backup-restore 
> >> as the only mechanism for migrating my old data to the new 
> installation. 
> >> This is the only way I would be confident that I weren't screwing 
> >> anything up. 
> >> 
> >> 
> > Thank you very much for helping me out on this, awokd and Andrew. 
> Currently 
> > I'm leaning toward taking the safe path. If I understand correctly that 
> > means: 
> > 
> >    1. Backup everything that's on the SSD *and* the external storage 
> pool 
> >    HDDs - this will take a lot of time and space but that's the price I 
> have 
> >    to pay for the safety I get 
> >    2. Connect the new SSD, wipe the external drives 
> >    3. Install Qubes OS on the new SSD 
> >    4. Create external storage pools on the additional HDDs 
> >    5. Make the SSD the default pool; restore VMs for SSD 
> >    6. Make external disk 1 the default pool; restore VMs for this pool 
> >    7. Make external disk 2 the default pool; restore VMs for this pool 
> >    8. Switch default pool back to SSD 
> >    9. Done 
> > 
> > How does this sound? 
> > 
>
> I haven't personally used external storage pools, so I can't comment 
> there. (Thankfully, though, others have already weighed in on that part.) 
>
> Related to Brendan's point, in step 1, it's important to *verify* the 
> backups you've just created. 
>
> GUI: Qube Manager -> Restore qubes from backup -> [x] Verify backup 
>      integrity, do not restore the data 
>
> CLI: qvm-backup-restore --verify-only [...] 
>
> As the saying goes: Backups always succeed. It's restores that fail. 
>
> (If you have the extra disks, it would technically be even safer to 
> use new HDDs and refrain from wiping the existing ones until after 
> you've verified that everything is correct in the new system, but this 
> might be overly cautious. I personally don't feel the need to do this 
> anymore, because I know the data is already there in a verified Qubes 
> backup, and I've tested my ability to manually recover it 
> independently of Qubes as a last resort.) 
>
> Aside from these caveats, your plan sounds like what I would do. 
>
> - -- 
> Andrew David Wong (Axon) 
> Community Manager, Qubes OS 
> https://www.qubes-os.org 
>
> -----BEGIN PGP SIGNATURE----- 
>
> iQIzBAEBCgAdFiEEZQ7rCYX0j3henGH1203TvDlQMDAFAl1sloEACgkQ203TvDlQ 
> MDDqdg//cFoY4k/EFm5t2pVpoiFKubh/o34CzJo8eYfOWgHaNnnVlhfqXGartWkS 
> F5aSeig2PiPWpPqfJ9G4mWV46ySjC/hez0fSmAl2rsNA/PaRTAG/aIhIy7DlNuoo 
> H9MQJw5TsiHvgz6h/FfYVOcl1/mDOwmVw4n9WQ1x49bgsmI+CdZf94c7P+XvEpde 
> YM3G2hGi6e1Bd3H3Xm1B5EtovsMXC3ieDkXDYlun814t1jBv0BmVib93vRaltHIt 
> Bdd766BAvhkdMOOWONHfmOrw1An7FBm1uKIxdJb71w5Kltv8VV1S7xXq387/QmGF 
> fcdJljdEtArgxg4Pe35i03GseDjRfw1rhsH3TL/8PDjY2P1n+lwI5JG8+fa7ZPUM 
> FHKoQAiPk9D3d8vOXmnwVT8QrCdaJvnX0bqtihusUnUh13+ZCrP5akq3KE7hrIn3 
> dgVG6eywbwS/Y18pbXChdvDdz3kx6Q05cB56nsFfyR65amLJh7F3NmkVd20HBDoh 
> yNxEqWMaHLiz/chOLLXFxUy8nAor/CQ8JgRPbERh40M5l67jzhFEXgHRl5u4XmbM 
> g8iNpYbFoUsBDP8bSzgPIFaNJ/OuUGnNsXtyYGwfxTzH45UMHLqGCqAPPdRUqaRB 
> W3JYH81cFnRVJdqBU+bj+1GPyD6an31++0ahJFX11DawHiP8nEc= 
> =d7uO 
> -----END PGP SIGNATURE----- 
>
>
Here is an update on how my migration from SSD_small to SSD_big is going so 
far.

Just as a remindet this is the challenge I face:
* dom0 SSD has 100 GB capacity, ~10% of this is free (that's why I want to 
migrate to a new SSD)
* external storage pool 1 has 1 TB storage, AppVM *1* with < 500 GB private 
storage in use
* external storage pool 2 has 1 TB storage, AppVM *2* with > 500 GB private 
storage in use
* I want to migrate everything via backup+restore to new disks/pools

*Here is what worked*
* backing up App VMs from all 3 pools using built-in backup mechanisms (UI) 
- cool

*Here is what did not work*
* *verifying* the huge (400-700 GB) backups *did not work* since this 
filled up my dom0 pretty fast and then failed -> this is the reason why I 
resorted to what Andrew wrote: having the original still in place while 
restoring to different disks, not overwriting anything, just in case 
restoring fails
* *restoring* the huge (400-700 GB) backups *did not work* since this 
filled up my dom0 pretty fast and then failed -> this is exactly like 
donoban wrote; I managed to work around this for AppVM *1*, NOT for AppVM 
*2* (yet)

To restore AppVM *1* (< 500 GB) I modified *restore.py 
<https://github.com/QubesOS/qubes-core-admin-client/blob/9158412a24da300e4c54346ccb54fce1e748500f/qubesadmin/backup/restore.py#L858>*
 
to restore to another location than */var/tmp*. The easiest for me was to 
create a new (temporary) AppVM in my new 1 TB external storate pool *1*, to 
increase its private storage to 500 GB, to mount its private volume to dom0 
and to use this path as temporary location in *restore.py*. So I was using 
my 1 TB disk both as restore target and temporary location for backup 
extraction. I was lucky - the pool filled up to 99.8% and the restore 
succeeded. So currently it seems you need double the amount of storage your 
to-be-restored AppVM consumes to restore the AppVM.

Now there is one challenge left. I have to restore AppVM *2* which is about 
700 GB. To my current knowledge I would now need to have twice this amount 
to restore - which currently I don't have. This is why I'd like to somehow 
slow down the extraction. donoban mentioned this is possible. I had a look 
at restore.py 
<https://github.com/QubesOS/qubes-core-admin-client/blob/master/qubesadmin/backup/restore.py>
 
but honestly have not idea where to start. I also currently don't know how 
the different extraction processes interact and how the backup is 
structured.

Can anybody suggest a modification (or hack, however dirty - it's meant to 
be temporary) to restore.py so it won't need 700 GB of additional temporary 
storage when I try to restore my 700 GB AppVM?

Thanks for all your input so far. Knowing that dom0 could fill up certainly 
saved my some hours of questioning life.

-- 
You received this message because you are subscribed to the Google Groups 
"qubes-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to qubes-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/qubes-users/90ead0e1-bdbe-4079-8b53-78042a4ffd8a%40googlegroups.com.

Reply via email to