Bug 1476343

Description of problem:

The autofs package that was released with RHEL 6.8 (autofs-5.0.5-122) worked properly over an stunnel connection.

The release shipped with RHEL 6.9 (5.0.5-132) prevents any connection from happening to the local stunnel process and breaks mounts to remote systems over stunnel connections.

Version-Release number of selected component (if applicable):

5.0.5-132

How reproducible:

There is a reliable reproducer.

Steps to Reproduce:

1) Set up a server/client stunnel connection targeted to port 2049
2) Set up an NFSv4 mount on the NFS server that binds to stunnel
3) Set up an NFSv4 automount on the NFS client that binds to stunnel
4) Attempt to mount the autofs location via navigating to the directory

A sample working environment with a failing test can be generated by doing the following:

1) Set up RVM (https://rvm.io/) and install Ruby version 2.1.9 and bundler
2) Run `git clone https://github.com/simp/pupmod-simp-simp_nfs
3) Run 'cd pupmod-simp-simp_nfs'
4) Run 'bundle update'
5) Ensure that VirtualBox is installed and properly functioning and has space to spawn around 4 systems
6) Run 'BEAKER_destroy=onpass bundle exec rake beaker:suites[stunnel]'
7) On failure, navigate to './vagrant/beaker_vagrant_files/default.yml'
8) Run 'vagrant ssh client6 -- -l root' in two different windows
9) Run 'service autofs stop'
10) Run 'automount -fvd'
11) Run 'cd /home/test.user' in the other window and watch the output of 'automount'

Actual results:

The following error message is output by 'automount -fvd':

[root@client6 etc]# automount -f -vd
Starting automounter version 5.0.5-132.el6, master map auto.master
using kernel protocol version 5.02
lookup_nss_read_master: reading master files auto.master
do_init: parse(sun): init gathered global options: (null)
lookup_read_master: lookup(file): read entry /home
master_do_mount: mounting /home
automount_path_to_fifo: fifo name /var/run/autofs.fifo-home
lookup_nss_read_map: reading map file /etc/autofs/home.map
do_init: parse(sun): init gathered global options: (null)
mounted indirect on /home with timeout 300, freq 75 seconds
st_ready: st_ready(): state = 0 path /home
handle_packet: type = 3
handle_packet_missing_indirect: token 194, name test.user, request pid 24290
attempting to mount entry /home/test.user
lookup_mount: lookup(file): looking up test.user
lookup_mount: lookup(file): test.user -> -fstype=nfs4,port=2049,sec=sys 127.0.0.1:/home/&
parse_mount: parse(sun): expanded entry: -fstype=nfs4,port=2049,sec=sys 127.0.0.1:/home/test.user
parse_mount: parse(sun): gathered options: fstype=nfs4,port=2049,sec=sys
parse_mount: parse(sun): dequote("127.0.0.1:/home/test.user") -> 127.0.0.1:/home/test.user
parse_mount: parse(sun): core of entry: options=fstype=nfs4,port=2049,sec=sys, loc=127.0.0.1:/home/test.user
sun_mount: parse(sun): mounting root /home, mountpoint test.user, what 127.0.0.1:/home/test.user, fstype nfs4, options port=2049,sec=sys
mount_mount: mount(nfs): root=/home name=test.user what=127.0.0.1:/home/test.user, fstype=nfs4, options=port=2049,sec=sys
mount_mount: mount(nfs): nfs options="port=2049,sec=sys", nobind=0, nosymlink=0, ro=0
mount_mount: mount(nfs): calling mkdir_path /home/test.user
mount(nfs): nfs: mount failure 127.0.0.1:/home/test.user on /home/test.user
dev_ioctl_send_fail: token = 194
failed to mount /home/test.user
handle_packet: type = 3
handle_packet_missing_indirect: token 195, name test.user, request pid 24290
dev_ioctl_send_fail: token = 195
statemachine:1472: got unexpected signal 28!

Expected results:

Normal expected behavior without errors. 

Additional info:

This has been filed as a CentOS bug here: 

https://bugs.centos.org/view.php?id=13575

The customer who filed that bug has indicated that the bug is reproducible in RHEL as well.