Xen Orchestra - Permissions Issue - File level restore

Danp

There was definitely something wrong with the VM's state. Whoami returned unknown uid. Had to forcibly reboot from within XO.

Danp

Was actually able to pull up a file listing from one of my backups. Here's what showed in the log -

Dec 21 12:07:02 xo-5 ntfs-3g[1857]: Version 2015.3.14AR.1 integrated FUSE 28
Dec 21 12:07:02 xo-5 ntfs-3g[1857]: Mounted /dev/loop1 (Read-Only, label "", NTFS 3.1)
Dec 21 12:07:02 xo-5 ntfs-3g[1857]: Cmdline options: ro
Dec 21 12:07:02 xo-5 ntfs-3g[1857]: Mount options: ro,allow_other,nonempty,relatime,fsname=/dev/loop1,blkdev,blksize=4096
Dec 21 12:07:02 xo-5 ntfs-3g[1857]: Ownership and permissions disabled, configuration type 7
Dec 21 12:07:02 xo-5 ntfs-3g[1857]: Unmounting /dev/loop1 ()

Danp

@DustinB3403 said in Xen Orchestra - Permissions Issue - File level restore:

Yeah I'm still getting that fusermount error...

What is the output from whereis fusermount?

Danp

@DustinB3403 said in Xen Orchestra - Permissions Issue - File level restore:

So are you able to restore files?

Yes, it appears to work in those cases where I can actually get a file listing.

DustinB3403

@Danp said in Xen Orchestra - Permissions Issue - File level restore:

@DustinB3403 said in Xen Orchestra - Permissions Issue - File level restore:

Yeah I'm still getting that fusermount error...

What is the output from whereis fusermount?

xoadmin@xo-test-debian-3:~$ whereis fusermount
fusermount:

DustinB3403

I do have fusermount under

/usr/share/bash-completion/completions/fusermount

Danp

This is what I get

xoadmin@xo-5:~$ whereis fusermount
fusermount: /bin/fusermount /usr/share/man/man1/fusermount.1.gz

Guessing that you either don't have it properly installed or it isn't in a location where it can be found.

DustinB3403

@Danp Do you have SSHFS installed?

DustinB3403

One sec, Fuse not installed on debian by default... FML... repo added installing fuse.

Rebooting and then recompiling..

Danp

@DustinB3403 Thought you did that yesterday. Did you rebuild from scratch?

DustinB3403

@Danp said in Xen Orchestra - Permissions Issue - File level restore:

@DustinB3403 Thought you did that yesterday. Did you rebuild from scratch?

No I added the universal repo yesterday for something else that was needed.

Anyways, now trying to mount a single backup I get

Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: Wed, 21 Dec 2016 19:15:18 GMT xo:api [email protected] | backup.scanDisk(...) [1s] ==> object
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: { [Error: Command failed: mount --options=loop,ro --source=/tmp/tmp-416paluKS1uHkUS/vhdi7 --target=/tmp/tmp-4160yW3G6ma9LZP
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: mount: wrong fs type, bad option, bad superblock on /dev/loop0,
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: missing codepage or helper program, or other error
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: In some cases useful info is found in syslog - try
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: dmesg | tail or so.
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: ]
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: code: 32,
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: killed: false,
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: stdout: '',
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: stderr: 'mount: wrong fs type, bad option, bad superblock on /dev/loop0,\n       missing codepage or helper program, or other error\n\n       In some cases useful info is found in syslog - try\n       dmesg | tail or so.\n',
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: failed: true,
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: signal: null,
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: cmd: 'mount --options=loop,ro --source=/tmp/tmp-416paluKS1uHkUS/vhdi7 --target=/tmp/tmp-4160yW3G6ma9LZP' }
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: Wed, 21 Dec 2016 19:15:18 GMT xo:api [email protected] | backup.scanFiles(...) [218ms] =!> Error: Command failed: mount --options=loop,ro --source=/tmp/tmp-416paluKS1uHkUS/vhdi7 --target=/tmp/tmp-4160yW3G6ma9LZP
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: mount: wrong fs type, bad option, bad superblock on /dev/loop0,
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: missing codepage or helper program, or other error
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: In some cases useful info is found in syslog - try
Dec 21 14:15:18 xo-test-debian-3 xo-server[416]: dmesg | tail or so.

Trying another backup . oh and dmesg | tail contains

10.867971] Adding 392188k swap on /dev/xvda5.  Priority:-1 extents:1 across:392188k SSFS
13.590095] RPC: Registered named UNIX socket transport module.
13.590099] RPC: Registered udp transport module.
13.590100] RPC: Registered tcp transport module.
13.590101] RPC: Registered tcp NFSv4.1 backchannel transport module.
13.633648] FS-Cache: Loaded
13.681056] FS-Cache: Netfs 'nfs' registered for caching
13.943515] Installing knfsd (copyright (C) 1996 [email protected]).
86.204993] fuse init (API version 7.23)
86.635229] loop: module loaded

DustinB3403

Seems to be specific to that backup, as I am able to restore files to my file server now without issue.

So a few things to we should try to compile into the update script, check for fuse, and install it if its not there.

DustinB3403

Another is listing things like this....

398.675872] ntfs: (device loop0): ntfs_readdir(): Directory index record with vcn 0x0 is corrupt.  Corrupt inode 0xe96b9.  Run chkdsk.
398.675890] ntfs: (device loop0): ntfs_readdir(): Directory index record with vcn 0x0 is corrupt.  Corrupt inode 0xe96b9.  Run chkdsk.

So yeah.. gonna run a check disk on my VM here..

Danp

@DustinB3403 said in Xen Orchestra - Permissions Issue - File level restore:

Seems to be specific to that backup, as I am able to restore files to my file server now without issue.

So a few things to we should try to compile into the update script, check for fuse, and install it if its not there.

Also for libvhdi-utils

DustinB3403

@Danp That was the other thing, that I installed yesterday.

libvhdi-utils

DustinB3403

This would list if the packaged is installed, so how to pipe that output into the next part would be the question.

dpkg --get-selections | grep libvhdi-utils
libvhdi-utils                                     install

Danp

Found this code on the interweb. I would think either the 2nd or 3rd solution would work fine for our needs.

DustinB3403

Yeah, this might be the easiest solution...

if [ $(dpkg-query -W -f='${Status}' nano 2>/dev/null | grep -c "ok installed") -eq 0 ];
then
  apt-get install nano;
fi

Might need to throw up a flag or warning if the package can't be found though. Like in my case I had to add the universal repo to my debian installation before I could install libvhdi-util and fuse.

Danp

Should be able to check the exit code from apt-get. Now to wrap the whole thing into a nice little function that we can reuse in the future.

Danp

5.5.2 is out. Grab it while it's hot!