Bug 1376819

Problem is because of behavior of "simple" unit type. Solution seems to be to migrate to notify. It's 7.4 material.

Created attachment 1545733 [details]
qnetd: Check existence of NSS DB dir before fork

qnetd: Check existence of NSS DB dir before fork

Previously, when user tried start corosync-qnetd without
initialized NSS database then generic (not very helpful
and misleading) NSS error was logged
"NSS error (-8015): The certificate/key database is in an old,
unsupported format.".

Solution is to check if it's possible to open NSS DB directory and
display (usually much more informative) result of strerror function.

Such check is called before fork, so init system can return error code
during start.

To make error reporting work with systemd it's also needed to change
unit type from simple to forking.

Signed-off-by: Jan Friesse <jfriesse>
Reviewed-by: Christine Caulfield <ccaulfie>

For QA: Patch solves only described scenario when NSS DB doesn't exists (what is probably most common reason of failure). All other "failures" are handled after fork so behavior is same as before.

What I've tested:
# ls -la /etc/corosync/qnetd 
ls: cannot access /etc/corosync/qnetd: No such file or directory

# systemctl start corosync-qnetd; echo $?
Job for corosync-qnetd.service failed because the control process exited with error code. See "systemctl status corosync-qnetd.service" and "journalctl -xe" for details.
1

# journalctl  _COMM=corosync-qnetd
* corosync-qnetd[*]: Can't open NSS DB directory (2): No such file or directory

After creating of CA:
# systemctl start corosync-qnetd; echo $?
0

# systemctl status corosync-qnetd; echo $?
● corosync-qnetd.service - Corosync Qdevice Network daemon
   Loaded: loaded (/usr/lib/systemd/system/corosync-qnetd.service; disabled; vendor preset: disabled)
   Active: active (running) since Tue 2019-03-19 16:51:27 CET; 2s ago
     Docs: man:corosync-qnetd
  Process: 17632 ExecStart=/usr/bin/corosync-qnetd $COROSYNC_QNETD_OPTIONS (code=exited, status=0/SUCCESS)
 Main PID: 17633 (corosync-qnetd)
...
0

Please note that error message is now (hopefully) more understandable - "No such file or directory" vs "NSS error (-8015): The certificate/key database is in an old, unsupported format.".

Created attachment 1546365 [details]
Use RuntimeDirectory instead of tmpfiles.d

Use RuntimeDirectory instead of tmpfiles.d
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

This reverts part of commit 32123f6bb2ebc4f9ac7865945cc85a9c9b903dc6.

A simple directive is a much lighter solution to the same problem, and
automatically follows the specified User.  I copied the 0770 modes from
the corresponding init scripts; they could use a little documentation.

Signed-off-by: Ferenc Wágner <wferi>
Reviewed-by: Jan Friesse <jfriesse>
(cherry picked from commit c733e9417ef1d2f31268e9b6f99a8fc7712fcea7)

Created attachment 1546367 [details]
configure: add --with-initconfigdir option

configure: add --with-initconfigdir option
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Default value is /etc/sysconfig and resulting
INITCONFIGDIR is used to reduce duplication in init system
integration code.

Signed-off-by: Ferenc Wágner <wferi>
Reviewed-by: Jan Friesse <jfriesse>
(cherry picked from commit d7208e88370d2bce40b45224a3971eeb68c22d3c)

BEFORE (corosync-qnetd-2.4.0-4.el7)
======
## Checks, that NSS DB doesn't exist
root@host-027 ~]# ls -l /etc/corosync/qnetd
ls: cannot access /etc/corosync/qnetd: No such file or directory

## Start the service without NSS DB, which should fail
[root@host-027 ~]# systemctl start corosync-qnetd.service
[root@host-027 ~]# echo $?
0

## systemctl shows status correctly
[root@host-027 ~]# systemctl status corosync-qnetd.service
● corosync-qnetd.service - Corosync Qdevice Network daemon
   Loaded: loaded (/usr/lib/systemd/system/corosync-qnetd.service; disabled; vendor preset: disabled)
   Active: failed (Result: exit-code) since Fri 2019-06-07 01:58:57 CDT; 2s ago
  Process: 8941 ExecStart=/usr/bin/corosync-qnetd -f $COROSYNC_QNETD_OPTIONS (code=exited, status=1/FAILURE)
 Main PID: 8941 (code=exited, status=1/FAILURE)

Jun 07 01:58:57 host-027.virt.lab.msp.redhat.com systemd[1]: Started Corosync Qdevice Network daemon.
Jun 07 01:58:57 host-027.virt.lab.msp.redhat.com systemd[1]: corosync-qnetd.service: main process ex...RE
Jun 07 01:58:57 host-027.virt.lab.msp.redhat.com systemd[1]: Unit corosync-qnetd.service entered fai...e.
Jun 07 01:58:57 host-027.virt.lab.msp.redhat.com systemd[1]: corosync-qnetd.service failed.
Hint: Some lines were ellipsized, use -l to show in full.

AFTER (corosync-qnetd-2.4.3-6.el7)
=====
## NSS DB doesn't exist
root@host-027 ~]# ls -l /etc/corosync/qnetd
ls: cannot access /etc/corosync/qnetd: No such file or directory

## systemctl start is now failing, as it should
[root@host-027 ~]# systemctl start corosync-qnetd.service
Job for corosync-qnetd.service failed because the control process exited with error code. See "systemctl status corosync-qnetd.service" and "journalctl -xe" for details.
[root@host-027 ~]# echo $?
1
[root@host-027 ~]# journalctl _COMM=corosync-qnetd
...
* corosync-qnetd[9809]: Can't open NSS DB directory (2): No such file or directory
...

RESULT
======
After the fix, corosync-qnetd service returns error status, when it is started without NSS DB.
Note: Both versions work fine when NSS DB is created
# corosync-qnetd-certutil -i

Verified for version corosync-2.4.3-6.el7

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2019:2245