[FACT-1992] autofs mountpoints being mounted by facter Created: 2019/08/16  Updated: 2019/10/11  Resolved: 2019/09/17

Status: Resolved
Project: Facter
Component/s: None
Affects Version/s: FACT 3.11.9
Fix Version/s: FACT 3.14.5, FACT 3.13.4, FACT 3.11.10

Type: Bug Priority: Major
Reporter: Fabrizio Lungo Assignee: Gabriel Nagy
Resolution: Fixed Votes: 0
Labels: resolved-issue-added
Remaining Estimate: 1 hour
Time Spent: Not Specified
Original Estimate: 1 hour

Issue Links:
is caused by FACT-1910 Mountpoint fact only shows tmpfs and ... Closed
Acceptance Criteria:

Running facter does not trigger autofs automounts to become mounted.

Team: Night's Watch
Story Points: 3
Sprint: NW - 2019-09-18
Method Found: Needs Assessment
Release Notes: Bug Fix
Release Notes Summary: When Facter stats a mountpoint to get the size and available space, it causes mountpoints of type {{autofs}} to be automatically mounted, which is not the intended behavior.

Automounts are now skipped by Facter when resolving mountpoints.
QA Risk Assessment: Needs Assessment


After updating to facter 3.11.9 we started experiencing issues caused by the regular mounting and unmounting of automount filesystems. We traced the cause back to facter and found that it was triggering a mount of . This seems to be related to the change made in FACT-1910 (0f0d8f4df70f886ec7ae86f65a24909a4f20b0f7/lib/src/facts/linux/filesystem_resolver.cc). The auto mount is triggered by stating the mount dir to get the size and available space.

Skipping non-physical mountpoints was a good catch all of a lot of synthetic filesystems which could introduce other problems problems and so it seems it will now be required to maintain a list of mount types to ignore.

The change should be reverted, autofs should be added as a skipped mtype (as with tempfs) or you should not stat autofs mountpoints.

Comment by Jean Bond [ 2019/10/10 ]

Gabriel Nagy, this went into Facter 3.14.5 as well, didn't it?

Comment by Gabriel Nagy [ 2019/10/10 ]

Yep, it did

Comment by Gabriel Nagy [ 2019/10/10 ]

Also 3.13.4 and 3.11.10 from the looks of the commit

Comment by Jean Bond [ 2019/10/10 ]

Oh, good, thank you Gabriel Nagy! I thought that's what the .z fixversions referred to, but with a .z version, I'm never sure.

Generated at Sun Jul 12 23:17:53 PDT 2020 using Jira 8.5.2#805002-sha1:a66f9354b9e12ac788984e5d84669c903a370049.