[Bug 2641] Add systemd notify code to to track running server

[bugzilla-daemon at bugzilla.mindrot.org]

https://bugzilla.mindrot.org/show_bug.cgi?id=2641

Michal Koutný <mkoutny at suse.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mkoutny at suse.com

--- Comment #10 from Michal Koutný <mkoutny at suse.com> ---
Created attachment 3099
  --> https://bugzilla.mindrot.org/attachment.cgi?id=3099&action=edit
notify systemd when daemon reloading

I observed a race similar to restarts during SSH daemon reloads.

Environment:
SLES12 SP2
openssh-7.2p2-74.1
systemd-228-150.15.2

Reproducer:
ssh $REMOTE_HOST systemctl reload sshd.service ; ssh $REMOTE_HOST

Actual behavior:
The second SSH connection fails in roughly 50% cases.

Expected behavior:
First SSH blocks until daemon reloads and second connection succeeds.

I propose to extend the suggested patch in comment 6 with another
sd_notify() call (attached) so that systemd can properly track
reloading state of the daemon.

The two patches fixed the behavior in my testcase (I admit I didn't
check actual systemd handling of the notification).

-- 
You are receiving this mail because:
You are watching someone on the CC list of the bug.
You are watching the assignee of the bug.