gc cannot destroy snapshot due to 2 pools being mounted as home
sylvainfaivre opened this issue · 3 comments
Describe the bug
Fully updated Ubuntu 22.10 machine. First installed with Ubuntu 21.10.
My setup is fully encrypted.
Trying to tun gc :
❯ sudo zsysctl service gc
ERROR Couldn't fully destroy user state rpool/USERDATA/sylvain_loeo0c: Couldn't destroy rpool/USERDATA/sylvain_loeo0c: couldn't destroy "rpool/USERDATA/sylvain_loeo0c" and its children: cannot destroy dataset "rpool/USERDATA/sylvain_loeo0c": dataset is busy.
Putting it in keep list.
ERROR Couldn't fully destroy user state rpool/USERDATA/sylvain_qn532u@autozsys_qjwbla: Couldn't destroy rpool/USERDATA/sylvain_qn532u@autozsys_qjwbla: couldn't destroy "rpool/USERDATA/sylvain_qn532u@autozsys_qjwbla" due to clones: "rpool/USERDATA/sylvain_qn532u@autozsys_qjwbla" has some clones ([rpool/USERDATA/sylvain_loeo0c]) when it shouldn't.
Putting it in keep list.
List all snapshots on this pool :
❯ sudo zfs list -rt all rpool/USERDATA
NAME USED AVAIL REFER MOUNTPOINT
rpool/USERDATA 151G 16.2G 192K /
rpool/USERDATA/root_qn532u 4.59G 16.2G 4.59G /root
rpool/USERDATA/root_qn532u@autozsys_ktrxrn 320K - 4.59G -
rpool/USERDATA/root_qn532u@autozsys_v2erx1 96K - 4.59G -
rpool/USERDATA/root_qn532u@autozsys_9q50s0 280K - 4.59G -
rpool/USERDATA/sylvain_loeo0c 41.4G 16.2G 124G /home/sylvain
rpool/USERDATA/sylvain_qn532u 105G 16.2G 105G /home/sylvain
rpool/USERDATA/sylvain_qn532u@autozsys_qjwbla 3.96M - 105G -
rpool/USERDATA/sylvain_qn532u@autozsys_ktrxrn 0B - 105G -
rpool/USERDATA/sylvain_qn532u@autozsys_v2erx1 0B - 105G -
rpool/USERDATA/sylvain_qn532u@autozsys_9q50s0 0B - 105G -
rpool/USERDATA/sylvain_qn532u@autozsys_ga1sja 0B - 105G -
rpool/USERDATA/sylvain_qn532u@autozsys_j9uroz 0B - 105G -
rpool/USERDATA/sylvain_qn532u@autozsys_3929kg 0B - 105G -
As you can see, 2 volumes are mounted as my home :
❯ mount |grep home
rpool/USERDATA/sylvain_qn532u on /home/sylvain type zfs (rw,relatime,xattr,posixacl)
rpool/USERDATA/sylvain_loeo0c on /home/sylvain type zfs (rw,relatime,xattr,posixacl)
I'm not quite sure how I can check which one really contains my data, and how I can fix my setup so that only one shows as mounted.
Well, the _loeo0c one seems to be the remains from some old stuff, as it is smaller, and doesn't matches the name of any snapshot.
See also :
❯ sudo zsysctl show
Name: rpool/ROOT/ubuntu_bcuvug
ZSys: true
Last Used: current
History:
- Name: rpool/ROOT/ubuntu_bcuvug@autozsys_9q50s0
Created on: 2023-01-12 10:03:01
- Name: rpool/ROOT/ubuntu_bcuvug@autozsys_v2erx1
Created on: 2023-01-11 11:35:02
- Name: rpool/ROOT/ubuntu_bcuvug@autozsys_ktrxrn
Created on: 2023-01-10 11:55:15
Users:
- Name: root
History:
- rpool/USERDATA/root_qn532u@autozsys_9q50s0 (2023-01-12 10:03:01)
- rpool/USERDATA/root_qn532u@autozsys_v2erx1 (2023-01-11 11:35:02)
- rpool/USERDATA/root_qn532u@autozsys_ktrxrn (2023-01-10 11:55:15)
- Name: sylvain
History:
- rpool/USERDATA/sylvain_qn532u@autozsys_3929kg (2023-01-12 13:27:23)
- rpool/USERDATA/sylvain_qn532u@autozsys_j9uroz (2023-01-12 12:26:23)
- rpool/USERDATA/sylvain_qn532u@autozsys_ga1sja (2023-01-12 11:25:23)
- rpool/USERDATA/sylvain_qn532u@autozsys_9q50s0 (2023-01-12 10:03:01)
- rpool/USERDATA/sylvain_qn532u@autozsys_v2erx1 (2023-01-11 11:35:02)
- rpool/USERDATA/sylvain_qn532u@autozsys_ktrxrn (2023-01-10 11:55:15)
- rpool/USERDATA/sylvain_qn532u@autozsys_qjwbla (2022-05-31 10:28:55)
- rpool/USERDATA/sylvain_loeo0c (0001-01-01 00:00:00)
To Reproduce
I don't know how it got to this.
I just looked into this because I started getting warnings about zsys not creating new snapshots because free space was less than 20%.
Expected behavior
Be able to delete old snapshots.
For ubuntu users, please run and copy the following:
ubuntu-bug zsys --save=/tmp/report
Sorry, cannot paste :
There was an error creating your Issue: body is too long, body is too long (maximum is 65536 characters).
Screenshots
n/a
Installed versions:
- OS: Ubuntu 22.10
- Zsysd running version: zsysctl 0.5.9, zsysd 0.5.9
Additional context
I might have used grub option to boot into an older snaphshot, a few months ago.
I might also have used other zfs snapshot tools at some point.
Some more info that might be relevant :
root@sylvain-thinkpad:~# lsblk -f |grep -v '^loop'
NAME FSTYPE FSVER LABEL UUID FSAVAIL FSUSE% MOUNTPOINTS
zd0 crypto_LUKS 2 74cb1202-c3fd-4cc2-b6e8-371de9809bd7
└─keystore-rpool ext4 1.0 keystore-rpool 036e9411-dc83-4357-b9c8-03878ea4bf2e 418,2M 0% /run/keystore/rpool
zd16 swap 1 2bd9bcac-158c-489f-bd8d-e96410ce512b [SWAP]
nvme0n1
├─nvme0n1p1 vfat FAT32 50B8-61AB 496,8M 3% /boot/grub
│ /boot/efi
├─nvme0n1p2
│ └─cryptoswap swap 1 cryptoswap f31ae71c-1c8f-499b-a477-d3b86d429d0c [SWAP]
├─nvme0n1p3 zfs_member 5000 bpool 17336107262982257653
└─nvme0n1p4 zfs_member 5000 rpool 4099056407353820615
root@sylvain-thinkpad:~# parted /dev/zd0 print
Error: /dev/zd0: unrecognised disk label
Model: Unknown (unknown)
Disk /dev/zd0: 524MB
Sector size (logical/physical): 512B/8192B
Partition Table: unknown
Disk Flags:
root@sylvain-thinkpad:~# parted /dev/nvme0n1 print
Model: SSSTC CA5-8D256-Q79 (nvme)
Disk /dev/nvme0n1: 256GB
Sector size (logical/physical): 512B/512B
Partition Table: gpt
Disk Flags:
Number Start End Size File system Name Flags
1 1049kB 538MB 537MB fat32 EFI System Partition boot, esp
2 538MB 2685MB 2147MB swap
3 2685MB 4833MB 2147MB zfs
4 4833MB 256GB 251GB zfs
Well I really shot myself in the foot here...
- I switched to another user account
- I checked which filesystem contained up-to-date data : it was
rpool/USERDATA/sylvain_loeo0c - I checked the contents of all autozsys snapshots for
rpool/USERDATA/sylvain_qn532u: they all had the same data from may 2022, so zsys had been doing snapshots of an older version of the filesystem for all these months - Then, instead of manually deleting the older filesystem and snapshots, I ran
zsysctl service gc, which promptly destroyed my current data
I'll take this as a learning opportunity, at least I still have my files from may 2022 (and some backups too).
Now I'm gonna close this bug as it's a dupe for #218