Backport Of Stretch’s Asterisk.service File Into Jessie: Successful Start Not Detected By Systemd
Home » Asterisk Users » Backport Of Stretch’s Asterisk.service File Into Jessie: Successful Start Not Detected By Systemd
Hello,
I’ve been tasked to enable automatic Asterisk restart on failure on a Jessie platform (running latest Asterisk 13.15.0).
I build a dedicated Jessie VM on which I installed Asterisk from source. I configured a couple of files in /etc/asterisk directory. I positively checedk that with simple config, Asterisk could sucessively start using an /etc/init.d/asterisk file and sysv-systemd compatibility tools.
Then I copied a /etc/systemd/system/asterisk.service file with the following content:
[Unit]
Description=Asterisk PBX
Documentation=man:asterisk(8)
Wants=network-online.target After=network-online.target
[Service]
Type=notify ExecStart=/usr/sbin/asterisk -g -f -U asterisk ExecReload=/usr/sbin/asterisk -rx ‘core reload’
Restart=no RestartSec=1
WorkingDirectory=/var/lib/asterisk
[Install]
WantedBy=multi-user.target
Please, note that this file has the same content as file
/lib/systemd/system/asterisk.service in Debian Stretch expect for line Restart=on-failure I changed to Restart=no, for the moment.
After creating above asterisk.service file, running “systemctl daemon-reload” and “systemctl start asterisk”, I could observe:
– asterisk is starting OK,
– asterisk prints to its log file a line such as:
[Apr 20 11:23:22] VERBOSE[770] asterisk.c: Asterisk Ready.
but, at the same time, I got:
april 20 11:23:22 jessievm asterisk[770]: [Apr 20 11:23:22] NOTICE[770]:
app_queue.c:9095 reload_queues: No call queueing config fil april 20 11:24:48 jessievm systemd[1]: asterisk.service start operation timed out. Terminating. april 20 11:24:48 jessievm asterisk[770]:
april 20 11:24:48 jessievm systemd[1]: Failed to start Asterisk PBX.
If I’m not mistaken:
– lastest “Asterisk Ready” line is visible in /var/log/asterisk/full but invisible in Systemd journal,
– things work as if Systemd was not notified of Asterisk successful start and than decided to stop Asterisk,
– man page [1] mentions support of type=notify in Debian Jessie
– on Stretch, I used asterisk 13.14.0 while I used asterisk 13.15.0, so I
supposed this issue doesn’t come from Asterisk itself but the way I
configured my system.
Systemd versions are:
Jessie:
systemd 215
+PAM +AUDIT +SELINUX +IMA +SYSVINIT +LIBCRYPTSETUP +GCRYPT +ACL +XZ
-SECCOMP -APPARMOR
Stretch:
systemd 232
+PAM +AUDIT +SELINUX +IMA +APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP
+GCRYPT +GNUTLS +ACL +XZ +LZ4 +SECCOMP +BLKID +ELFUTILS +KMOD +IDN
My questions are:
1. Is it possible to use systemd’s Restart=on-failure option to automaticaly restart Asterisk, given Jessie systemd version ?
2. If positive, how ?
3. If negative, which option would you recommend (Monit ? ) ?
Best regards
[1] https://manpages.debian.org/jessie/systemd/systemd.service.5.en.html
One thought on - Backport Of Stretch’s Asterisk.service File Into Jessie: Successful Start Not Detected By Systemd
[snip]
[snip]
‘Type=notify’ means that the service is only considered ready once is notifies systemd so through a special socket, using sd_notify(3). You told systemd that Asterisk will notify it when ready, but Asterisk never did. Thus systemd decided that service has failed.
Make sure you have libsystemd-dev installed. Alternatively, remove that line (to keep the default Type=simple).
When built with systemd support, asterisk updates systemd about its status using sd_notify(). This means that it is only considered as started when it has loaded all the modules. This can make startup ordering simpler.
BTW: if anybody wants, you can use sd_notify to set the status (as shown in ‘systemctl status asterisk.service’) to an arbitrary string. Basically just use ast_sd_notify() anywhere in the code. It becomes a no-op if there’s no systemd support.