TrueNAS 12 - NFS shares --> corrupt file systems

[tanjix]

Hello everyone,

I am having some trouble on my TrueNAS server (details see in signature spoiler).
Some NFS shares on TrueNAS are connectedvia an internal 10G link to my XCP-NG hosts, where VMs running on them.

I discovered some issues regarding to corrupt file systems (not sure if it's just on the NFS shares or on my iSCSI shares as well). Even if I create a new VM with let's say Debian 10 as guest os it's just a matter of time until when I get errors in dmesg about EXT4-fs errors, like:

Code:

[271756.735703] EXT4-fs warning (device xvda1): ext4_dirent_csum_verify:368: inode #1446954: comm iptables: No space for directory leaf checksum. Please run e2fsck -D.
[271756.735708] EXT4-fs error (device xvda1): ext4_dx_find_entry:1559: inode #1446954: block 1: comm iptables: Directory block failed checksum
[271756.736589] Aborting journal on device xvda1-8.
[271756.738252] EXT4-fs (xvda1): Remounting filesystem read-only
[271756.738851] EXT4-fs error (device xvda1): ext4_journal_check_start:61: Detected aborted journal
[271756.741602] EXT4-fs error (device xvda1): ext4_dx_find_entry:1559: inode #1446954: block 2: comm iptables: Directory block failed checksum
[271756.744547] EXT4-fs error (device xvda1): ext4_dx_find_entry:1559: inode #1446954: block 2: comm iptables: Directory block failed checksum
[271756.747187] EXT4-fs error (device xvda1): ext4_dx_find_entry:1559: inode #1446954: block 2: comm iptables: Directory block failed checksum
[271756.774957] EXT4-fs warning (device xvda1): ext4_dirent_csum_verify:368: inode #1446954: comm iptables: No space for directory leaf checksum. Please run e2fsck -D.
[271756.774961] EXT4-fs error (device xvda1): ext4_dx_find_entry:1559: inode #1446954: block 1: comm iptables: Directory block failed checksum

The server did mount in read-only, which is a normal behaviour here; however, after doing a reboot, a file system check is enforced and finds a lot of errors.

Could this a matter of conenction problems between the TrueNAS server and my XCP-NG hosts?
The disks within my TrueNAS are all in a good condition, no errors and the raid array itself is working properly as well.

Any advise would be highly appreciated as such an issue might easily get out of control by loosing lots of data.

Thanks!

 

[sretalla]

Could this a matter of conenction problems between the TrueNAS server and my XCP-NG hosts?

OS: FreeNAS 12.0-U1

This could be part of it... I think there was a bug which sounds a bit like what you're talking about here which was fixed in U2 or U2.1.

EDIT: actually it was U1.1:

/_archive/corereleasenotes/12.0/12.0u1.1/