So recently we’ve been seeing an interesting issue since updating a server from 23.10.2 to 24.10.2 version TrueNAS SCALE.
It seems to be causing an unacceptably high number of skipped files in our backups.
Outline of the problem we are seeing:
Backup jobs having high number of skipped files, common factor on the skipped files is they have Alternate Data Streams attribute set.
We are using Arcserve UDP backup off a Mixed NFS\SMB share accessed via SMB running on a Truenas SCALE 24.10.2 server.
The error seen in the backup software logs is: Failed to BackupRead(\?\UNC\storagenode.testdomain.co.uk\TestShare\TestFiles.txt),skip it.Ret:32 {AFBackend.exe::AFBackupVirtual.dll(6034.674)}
The normal operation of this share is by clients running Windows 10/11, Various Linux flavours and MacOS.
Share settings as follows, note ‘hide unreadable = Yes’ is being set via TruenasCLI tool to specify auxsmbconf parameters.
[TestShare]
access based share enum = Yes
ea support = No
hide unreadable = Yes
kernel oplocks = Yes
path = /mnt/tank0/Test_Share
read only = No
smbd max xattr size = 2097152
vfs objects = zfs_fsrvp streams_xattr shadow_copy_zfs ixnas zfs_core io_uring
tn:vuid = 744c36ff-a89c-404c-814e-4faa99681c74
fruit:time machine max size = 0
fruit:time machine = False
nfs4:chown = True
tn:home = False
tn:path_suffix =
tn:purpose = MULTI_PROTOCOL_NFS
Testing we’ve done:
Built test VM with 24.10.2 fresh install from media built the share, added some representative data. We’re able to reproduce the backup job issues skipping all the files with Alternate Data Streams.
Built test VM with 23.10.2 fresh install from media built the share, added some representative data. We were unable to recreate the issue. The backup ran fine for all files.
Upgraded the 23.10.2 test VM to 24.10.2 ran the backup job again and it is now exhibiting the skipped files issue.
Reading the release notes suggests there have been a number of modifications to the way Alternate Data Streams work between 23.10.2 and 24.10.2.
I’m sure I don’t truly understand the detail there or if this is related to my issue.
We are exploring the issue with Arcserve, but seeing as we can clearly replicate the fault by upgrading a test Truenas server from 23.10.2 to 24.10.2 thought someone here might have some ideas or already encountered the issue and knows how to deal with it.