Updated Mounting (markdown)

DatuX
2023-10-16 12:22:07 +02:00
parent cd6fecb7cf
commit 0f0374e737

@ -14,7 +14,7 @@ If you then proceed to use `--rollback` or `--force` things might get weird: ZF
**In version 3.3 of zfs-autobackup this will seem to be more prominent, because mounting has improved. Normally this problem would happen later after you reboot or zfs mount -a**
### Possible causes
## Possible causes
Consider this pool:
@ -30,7 +30,7 @@ This means the following mount-point directories where created:
* Dataset `pool` has a directory named `sub1`
* Dataset `pool/sub1` has a directory named `sub2`
#### Target side
### Target side
Now if we only select `pool` and `pool/sub1/sub2` to backup, then the target server will look like this:
@ -46,7 +46,7 @@ This means the following mount-point directories are needed:
So `sub2` will be created in `sub1`, thus modifying the dataset!
### Solution
## Solution
The best solution is to use the `--clear-mountpoint` option of zfs-autobackup. This will set canmount=noauto on newly received datasets.
@ -54,7 +54,7 @@ When you need to access the data you can use zfs mount to mount just one dataset
**Note that this option also prevents bootproblems, in case a dataset has a mountpoint that conflicts with existing mountpoints.**
#### Fix the problem manually
### Fix the problem manually
If you forgot to use --clear-mount, you can use some shell magic to fix it, for example for `test_target1`:
@ -65,7 +65,7 @@ If you forgot to use --clear-mount, you can use some shell magic to fix it, for
After this you will have to use the `--rollback` option once to remove all the changes.
### Workarounds
## Workarounds
These are some workarounds if cant use the method above, but they are not a perfect solution: