Error on puppet agent after upgrade to 3.7.1.

101 views
Skip to first unread message

Schofield

unread,
Nov 8, 2014, 3:51:19 PM11/8/14
to puppet...@googlegroups.com
After upgrading a puppet agent from v3.0.1 to v3.7.1 on a CentOS 6.5 server  I get the following error during a puppet agent run.

[junk-01: SU: lib/facter 455] puppet agent -t
Info: Retrieving pluginfacts
Info: Retrieving plugin
Info: Loading facts
/usr/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31: command not found: /etc/blkid /dev/sda1
/usr/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31: command not found: /etc/blkid /dev/sda2
/usr/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31: command not found: /etc/blkid /dev/sda3

Info: Caching catalog for junk-01

Any thoughts on the cause? I'm open to troubleshooting tips as well?


Denmat

unread,
Nov 8, 2014, 8:11:14 PM11/8/14
to puppet...@googlegroups.com
Hmmm. Not sure but here's what I would do..

I would ensure all rpm packages and any gems relating to puppet and facter are uninstalled, then install again from whatever package manager you prefer.

It feels like an unclean gem/rpm issue.

Den
--
You received this message because you are subscribed to the Google Groups "Puppet Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to puppet-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/puppet-users/aa24ff17-7d32-4921-9152-41edbdc48f42%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Schofield

unread,
Nov 10, 2014, 1:06:10 PM11/10/14
to puppet...@googlegroups.com
Isolated the problem.... Facter 2.1.0 introduced the partitions fact. This fact calls Facter::Core::Execution::Posix.which('blkid') to locate the blkid executable.  Turns out there is a directory /etc/blkid which unfortunately for us is on the PATH before /sbin/blkid.  Facter::Core::Execution::Posix.which() is incorrectly returning the directory instead of a true executable.

jcbollinger

unread,
Nov 11, 2014, 9:09:33 AM11/11/14
to puppet...@googlegroups.com


On Monday, November 10, 2014 12:06:10 PM UTC-6, Schofield wrote:
Isolated the problem.... Facter 2.1.0 introduced the partitions fact. This fact calls Facter::Core::Execution::Posix.which('blkid') to locate the blkid executable.  Turns out there is a directory /etc/blkid which unfortunately for us is on the PATH before /sbin/blkid.  Facter::Core::Execution::Posix.which() is incorrectly returning the directory instead of a true executable.


Facter appears to be buggy here, and I encourage you to file a ticket about it.  (It should ignore subdirectories in the executable search path, as the 'which' command it is trying to emulate does.)

In the meantime, however, it appears that to trigger the bug under your particular circumstances, you must have directory /etc in your path.  That's pretty non-standard; do you actually need it?  I think you could work around the issue by ensuring that it's not in your path.  At worst, you should be able to open a shell with a sanitized path, and launch the module tool from there.


John

Schofield

unread,
Nov 11, 2014, 1:24:57 PM11/11/14
to puppet...@googlegroups.com


On Tuesday, November 11, 2014 9:09:33 AM UTC-5, jcbollinger wrote:


Facter appears to be buggy here, and I encourage you to file a ticket about it.  (It should ignore subdirectories in the executable search path, as the 'which' command it is trying to emulate does.)

In the meantime, however, it appears that to trigger the bug under your particular circumstances, you must have directory /etc in your path.  That's pretty non-standard; do you actually need it?  I think you could work around the issue by ensuring that it's not in your path.  At worst, you should be able to open a shell with a sanitized path, and launch the module tool from there.

Agree, it's non-standard and shouldn't be needed.  Unfortunately it's required due to poor decisions over a decade ago.  Fortunately it doesn't appear to impact the puppet runs.



Thanks everyone for your input.
Reply all
Reply to author
Forward
0 new messages