Comprehensive data protection for all workloads
dandav
Influencer
Posts: 22 Liked: 4 times
Joined: Jan 15, 2021 2:53 am
Full Name: Daniel Davis
Contact:
Post
by dandav » Aug 24, 2021 12:03 am
this post
If you try and use XFS reflinks with the repository installed on a 5.10 Linux kernel VBR fails to detect the reflink support and returns the following error:
"File system behind the provided path has XFS reflinks diabled: /mnt/veeamrepo/backups Proceed anyway?"
On the Linux repository reflinks is enabled and I have tested it is working correctly using manuall file copies. In the VeeamTransportSvg.log it shows the below output which clearly shows from the xfs_info command reflinks is enabled on the filesystem. The only difference in the xfs_info output between kernels 5.4 and 5.10 is that the "bigtime = 0" variable doesn't exist in 5.4 so it appears as if this might have broken the code parsing the xfs_info output string.
Code: Select all
[24.08.2021 09:23:22.692] <139625863198464> cli | Method [FileSystem.GetXfsRefLinkSupport] completed. Elapsed time: [00:00:00.023]
[24.08.2021 09:23:22.692] <139625863198464> cli | - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[24.08.2021 09:23:22.692] <139625863198464> cli | Result
[24.08.2021 09:23:22.692] <139625863198464> cli | (EString) Answer = meta-data=/dev/sdb1 isize=512 agcount=30, agsize=268435455 blks
[24.08.2021 09:23:22.692] <139625863198464> cli | = sectsz=4096 attr=2, projid32bit=1
[24.08.2021 09:23:22.692] <139625863198464> cli | = crc=1 finobt=1, sparse=1, rmapbt=0
[24.08.2021 09:23:22.692] <139625863198464> cli | = reflink=1 bigtime=0
[24.08.2021 09:23:22.692] <139625863198464> cli | data = bsize=4096 blocks=8053063168, imaxpct=5
[24.08.2021 09:23:22.692] <139625863198464> cli | = sunit=0 swidth=0 blks
[24.08.2021 09:23:22.692] <139625863198464> cli | naming =version 2 bsize=4096 ascii-ci=0, ftype=1
[24.08.2021 09:23:22.692] <139625863198464> cli | log =internal log bsize=4096 blocks=521728, version=2
[24.08.2021 09:23:22.692] <139625863198464> cli | = sectsz=4096 sunit=1 blks, lazy-count=1
[24.08.2021 09:23:22.692] <139625863198464> cli | realtime =none extsz=4096 blocks=0, rtextents=0
[24.08.2021 09:23:22.692] <139625863198464> cli |
[24.08.2021 09:23:22.692] <139625863198464> cli | (EString) Command = xfs_info /mnt/veeamrepo
[24.08.2021 09:23:22.692] <139625863198464> cli | (EInt32) ExitCode = 0
[24.08.2021 09:23:22.692] <139625863198464> cli | (EString) StdError = <empty>
[24.08.2021 09:23:22.692] <139625863198464> cli | -------------------------------------------------------------------------------
HannesK
Product Manager
Posts: 14839 Liked: 3085 times
Joined: Sep 01, 2014 11:46 am
Full Name: Hannes Kasparick
Location: Austria
Contact:
Post
by HannesK » Aug 24, 2021 5:47 am
this post
Hello,
it's working fine with kernel 5.10 for me.
Code: Select all
root@kerneltest:~# uname -a
Linux kerneltest 5.10.59-051059-generic #202108151334 SMP Sun Aug 15 14:16:01 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux
do you have a Veeam support case number and more details about which distribution you are using?
Best regards,
Hannes
dandav
Influencer
Posts: 22 Liked: 4 times
Joined: Jan 15, 2021 2:53 am
Full Name: Daniel Davis
Contact:
Post
by dandav » Aug 24, 2021 7:01 am
this post
No, I hadn't logged a job as it wasn't a production issue, just playing in my lab.
Distribution is Debian 11.
Code: Select all
lhc@drrepo:/var/log/VeeamBackup$ uname -a
Linux drrepo 5.10.0-8-amd64 #1 SMP Debian 5.10.46-4 (2021-08-03) x86_64 GNU/Linux
dandav
Influencer
Posts: 22 Liked: 4 times
Joined: Jan 15, 2021 2:53 am
Full Name: Daniel Davis
Contact:
Post
by dandav » Aug 24, 2021 7:20 am
this post
Here's the comparison with our production Linux repository running Ubuntu 20.04. As you can see the xfs_info output is formatted identically except for the new bigtime variable on the same line as reflinks. Bigtime should show up in 5.10 kernels onwards as they now support 64-bit timestamps to fix the 2038 problem with 32 bit timestamps.
uname output:
Code: Select all
Linux veeamhl 5.4.0-81-generic #91-Ubuntu SMP Thu Jul 15 19:09:17 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux
VeeamTransportSvc.log output:
Code: Select all
[24.08.2021 16:36:30.616] <140665899947776> cli | Method [FileSystem.GetXfsRefLinkSupport] comp leted. Elapsed time: [00:00:00.019]
[24.08.2021 16:36:30.616] <140665899947776> cli | - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[24.08.2021 16:36:30.616] <140665899947776> cli | Result
[24.08.2021 16:36:30.616] <140665899947776> cli | (EString) Answer = meta-data=/dev/bcache0p1 isize=512 agcount=32, agsize=122090240 blks
[24.08.2021 16:36:30.616] <140665899947776> cli | = sectsz=4096 attr=2, projid32bit=1
[24.08.2021 16:36:30.616] <140665899947776> cli | = crc=1 finobt=1, sparse=1, rmapbt=0
[24.08.2021 16:36:30.616] <140665899947776> cli | = reflink=1
[24.08.2021 16:36:30.616] <140665899947776> cli | data = bsize=4096 blocks=3906885120, imaxpct=5
[24.08.2021 16:36:30.616] <140665899947776> cli | = sunit=128 swidth=512 blks
[24.08.2021 16:36:30.616] <140665899947776> cli | naming =version 2 bsize=4096 ascii-ci=0, ftype=1
[24.08.2021 16:36:30.616] <140665899947776> cli | log =internal log bsize=4096 blocks=521728, version=2
[24.08.2021 16:36:30.616] <140665899947776> cli | = sectsz=4096 sunit=1 blks, lazy-count=1
[24.08.2021 16:36:30.616] <140665899947776> cli | realtime =none extsz=4096 blocks=0, rtextents=0
[24.08.2021 16:36:30.616] <140665899947776> cli |
[24.08.2021 16:36:30.616] <140665899947776> cli | (EString) Command = xfs_info /mnt/veeamrepo 1
[24.08.2021 16:36:30.616] <140665899947776> cli | (EInt32) ExitCode = 0
[24.08.2021 16:36:30.616] <140665899947776> cli | (EString) StdError = <empty>
[24.08.2021 16:36:30.616] <140665899947776> cli | --------------------------------------------- ----------------------------------
HannesK
Product Manager
Posts: 14839 Liked: 3085 times
Joined: Sep 01, 2014 11:46 am
Full Name: Hannes Kasparick
Location: Austria
Contact:
Post
by HannesK » Aug 24, 2021 7:48 am
this post
my test was with Ubuntu 20.04 and kernel 5.10 and it worked fine.
Thanks for mentioning that you use Debian 11. I see the same issue. As for now
only Debian 10 is fully supported . So I'm pretty sure that QA will catch that issue before declaring full support for Debian 11. I will talk to them whether they already have it on their agenda.
UPDATE: planned to be supported in 11a (#312552)
dandav
Influencer
Posts: 22 Liked: 4 times
Joined: Jan 15, 2021 2:53 am
Full Name: Daniel Davis
Contact:
Post
by dandav » Aug 25, 2021 1:53 am
this post
Excellent, thanks Hannes
Users browsing this forum: Bing [Bot] , jfvm , veremin and 265 guests