I already put a ticket in, but I think an old bug ...
# compliance
r
I already put a ticket in, but I think an old bug has been reintroduced to SCE_Linux. Seems to only affect RHEL7 and 8 but not 9 and 10 This error pops up every agent run Error while resolving custom fact fact='sce_mount_info', resolution='<anonymous>': unexpected token at ''
m
I've a ticket in on this as well.
k
I also have a ticket for that, says "Resolved in v2.3.1 & v2.6.0". For us, it's caused by NFS mounts that aren't readable by root.
j
Did this start with the latest 2.6.0 that went out? There was a change in 2.6.0 specifically about
sce_mount_info
and I'm wondering if it resolves this issue or is causing this issue?
m
2.6.0 is currently the cause of this issue for me, with stig applied on RHEL7 it actually prevents catalog compilation.
r
Installed 2.6.0 and it started happening
j
Thank you! And thanks for making sure there is a ticket in for it. I'll drop in a message to clarify this is a new thing introduced by the change in 2.6.0 release that was pushed out.
h
@Mike Loseke @kenyon @RyChannel If y'all can, would y'all mind running the following commands on an affected machine and DM me with which ones throw errors and the sanitized error messages? I'm unable to reproduce the issue in our acceptance test environment, so I suspect it's due to `findmnt`'s interaction with remote filesystems. The commands are: •
findmnt -lmo SOURCE,TARGET,FSTYPE,FS-OPTIONS,FREQ,PASSNO,FSROOT,ID,UUID,LABEL,PARTUUID,PARTLABEL,MAJ:MIN,OPT-FIELDS,PARENT,PROPAGATION,TID,VFS-OPTIONS -J
•
findmnt -lso SOURCE,TARGET,FSTYPE,FS-OPTIONS,FREQ,PASSNO,FSROOT,ID,UUID,LABEL,PARTUUID,PARTLABEL,MAJ:MIN,OPT-FIELDS,PARENT,PROPAGATION,TID,VFS-OPTIONS -J
•
findmnt -lo SOURCE,TARGET,FSTYPE,FS-OPTIONS,FREQ,PASSNO,FSROOT,ID,UUID,LABEL,PARTUUID,PARTLABEL,MAJ:MIN,OPT-FIELDS,PARENT,PROPAGATION,TID,VFS-OPTIONS -J
m
findmnt: invalid option -- 'J'
is what I'm getting as an error.
And without -J:
findmnt: unknown column: PARENT,PROPAGATION,TID,VFS-OPTIONS
r
I can on Monday. We had to roll back wirh the weekend starting at noon
👍 1
h
@Mike Loseke was that on RHEL 7?
m
Yessir
findmnt provided by
util-linux-2.23.2-65.el7_9.1.x86_64
h
Thanks!
k
@Heston Snodgrass oh I'm still on sce_linux 2.5.0, so not seeing the error right now. If you still need more info to reproduce, let me know, I can try 2.6.0 on some machines
h
That would be awesome, thanks. I think I have it narrowed down, but I still can't repro on our test machines so the more logs I can look at the better.
k
ok, trying 2.6.0, which fixes some issues that we reported according to the release notes, so I need to try it anyawy. we use NetApp as the NFS server, I'm guessing you don't have that in your test env, you can probably configure a Linux NFS server the same way, but I don't know much about our server side config
r
Our RHEl 8 server that showed the error didn't have an NFS mount
h
The issue doesn't appear to be caused by mount types, or really anything related to parsing individual lines of the fstab file. The actual error is being thrown during saving the fstab file while comparing and ordering the file lines, which is odd. Once I get it completely figured out I can post the root cause for posterity. Sorry, that's relevant to a related bug, not this one. The
sce_mount_info
bug is being cause by how we are using
findmnt
and the changes between operating systems.
k
Well I upgraded production to 2.6.0, and am not seeing the error anywhere. My issue was Case# 01317452.