Gestió Servidors
2019-Nov-21 13:28 UTC
[CentOS] Issue with "ExecStartPost" attribute in systemd daemon faile
Hello, I'm trying to configure a daemon (I'm doing tests with "crond" daemon) to send me an email after daemon restart. My "crond.service" file is: # /etc/systemd/system/crond.service [Unit] Description=Command Scheduler After=auditd.service systemd-user-sessions.service time-sync.target #OnFailure=crond-notify-email@%i.service [Service] EnvironmentFile=/etc/sysconfig/crond ExecStart=/usr/sbin/crond -n $CRONDARGS ExecReload=/bin/kill -HUP $MAINPID KillMode=process Restart=on-failure RestartSec=10s ExecStartPost=/bin/sh -c "/etc/systemd/system/test.sh" [Install] WantedBy=multi-user.target My "test.sh" is very simple: #!/bin/sh echo "CROND is restarting" | /usr/bin/mailx -s "crond failure notification" myemailaddress at mydomain As you can see, I have added lines "Restart=on-failure", "RestartSec=10s" and " ExecStartPost=/bin/sh -c "/etc/systemd/system/test.sh"" to the original crond.service daemon file. Then, I run "systemctl daemon-reconfigure" and, from one console, I run "kill -9 `pidof cron`" for restarting crond daemon. After it, I receive an email... But now, if I run again ""kill -9 `pidof cron`", I don't receive any mail... I have notice that if I run "systemctl daemon-reload" and then kill crond process, mail is sent perfectly... but if I don't run "systemctl daemon-reload", mail is sent ONLY first time... Why? Thanks.
Leon Fauster
2019-Nov-21 13:41 UTC
[CentOS] Issue with "ExecStartPost" attribute in systemd daemon faile
Am 21.11.19 um 14:28 schrieb Gesti? Servidors:> Hello, > > I'm trying to configure a daemon (I'm doing tests with "crond" daemon) to send me an email after daemon restart. My "crond.service" file is: > # /etc/systemd/system/crond.service > [Unit] > Description=Command Scheduler > After=auditd.service systemd-user-sessions.service time-sync.target > #OnFailure=crond-notify-email@%i.service > > [Service] > EnvironmentFile=/etc/sysconfig/crond > ExecStart=/usr/sbin/crond -n $CRONDARGS > ExecReload=/bin/kill -HUP $MAINPID > KillMode=process > Restart=on-failure > RestartSec=10s > ExecStartPost=/bin/sh -c "/etc/systemd/system/test.sh" > > [Install] > WantedBy=multi-user.target > My "test.sh" is very simple: > #!/bin/sh > echo "CROND is restarting" | /usr/bin/mailx -s "crond failure notification" myemailaddress at mydomain > > As you can see, I have added lines "Restart=on-failure", "RestartSec=10s" and " ExecStartPost=/bin/sh -c "/etc/systemd/system/test.sh"" to the original crond.service daemon file. Then, I run "systemctl daemon-reconfigure" and, from one console, I run "kill -9 `pidof cron`" for restarting crond daemon. After it, I receive an email... But now, if I run again ""kill -9 `pidof cron`", I don't receive any mail... I have notice that if I run "systemctl daemon-reload" and then kill crond process, mail is sent perfectly... but if I don't run "systemctl daemon-reload", mail is sent ONLY first time... > > Why?What is "status" showing (systemctl status crond)? For both cases (first/second kill)? Any differences? -- Leon
James Pearson
2019-Nov-21 17:34 UTC
[CentOS] Issue with "ExecStartPost" attribute in systemd daemon faile
Gesti? Servidors wrote:> > Hello, > > I'm trying to configure a daemon (I'm doing tests with "crond" daemon) to send me an email after daemon restart. My "crond.service" file is: > # /etc/systemd/system/crond.service > [Unit] > Description=Command Scheduler > After=auditd.service systemd-user-sessions.service time-sync.target > #OnFailure=crond-notify-email@%i.service > > [Service] > EnvironmentFile=/etc/sysconfig/crond > ExecStart=/usr/sbin/crond -n $CRONDARGS > ExecReload=/bin/kill -HUP $MAINPID > KillMode=process > Restart=on-failure > RestartSec=10s > ExecStartPost=/bin/sh -c "/etc/systemd/system/test.sh" > > [Install] > WantedBy=multi-user.target > My "test.sh" is very simple: > #!/bin/sh > echo "CROND is restarting" | /usr/bin/mailx -s "crond failure notification" myemailaddress at mydomain > > As you can see, I have added lines "Restart=on-failure", "RestartSec=10s" and " ExecStartPost=/bin/sh -c "/etc/systemd/system/test.sh"" to the original crond.service daemon file. Then, I run "systemctl daemon-reconfigure" and, from one console, I run "kill -9 `pidof cron`" for restarting crond daemon. After it, I receive an email... But now, if I run again ""kill -9 `pidof cron`", I don't receive any mail... I have notice that if I run "systemctl daemon-reload" and then kill crond process, mail is sent perfectly... but if I don't run "systemctl daemon-reload", mail is sent ONLY first time... > > Why?I guess it is probably the way systemd works - subsequent 'crashes' (or kills in you case) probably only restart the main process (ExecStart) and don't re-run the ExecStartPost bits ... I notice you have a commented out an 'OnFailure' line in the unit file - so I guess you've been experimenting with that - which I believe is actually the way to do what you need ? James Pearson
Gestió Servidors
2019-Nov-22 10:07 UTC
[CentOS] Issue with "ExecStartPost" attribute in systemd daemon faile
Hellos,
Yes, I did some tests with "OnFailure" attribute, but the result was
the same: system didn't sent any mail.
Daemon and script files were:
* crond.service:
* [Unit]
Description=Command Scheduler
After=auditd.service systemd-user-sessions.service time-sync.target
OnFailure=crond-notify-email.service
[Service]
EnvironmentFile=/etc/sysconfig/crond
ExecStart=/usr/sbin/crond -n $CRONDARGS
ExecReload=/bin/kill -HUP $MAINPID
KillMode=process
Restart=on-failure
RestartSec=5s
[Install]
WantedBy=multi-user.target
* crond-notify-email.service:
* [Unit]
Description=crond failure email notification
After=crond.service
Requires=crond.service
[Service]
Type=oneshot
ExecStart=/etc/systemd/system/mail-notification.sh
* /etc/systemd/system/mail-notification.sh
* #!/bin/bash
echo "Crond restarted" | /usr/bin/mailx -s "Crond failure
notification" myemail at mydomain
After "systemctl daemon-reload" and "kill -9 `pidof crond`",
message log are:
Nov 22 11:04:36 mysystem systemd: crond.service: main process exited,
code=killed, status=9/KILL
Nov 22 11:04:36 mysystem systemd: Unit crond.service entered failed state.
Nov 22 11:04:36 mysystem systemd: Triggering OnFailure= dependencies of
crond.service.
Nov 22 11:04:36 mysystem systemd: crond.service failed.
Nov 22 11:04:41 mysystem systemd: crond.service holdoff time over, scheduling
restart.
Nov 22 11:04:41 mysystem systemd: Stopped Command Scheduler.
Nov 22 11:04:41 mysystem systemd: Started Command Scheduler.
Nov 22 11:04:41 mysystem systemd: Starting crond failure email notification...
Nov 22 11:04:41 mysystem systemd: Started crond failure email notification.
... but I don't receive any email...
Why?
El 21/11/2019 a las 18:34, James Pearson escribi?:
Gesti? Servidors wrote:
Hello,
I'm trying to configure a daemon (I'm doing tests with "crond"
daemon) to send me an email after daemon restart. My "crond.service"
file is:
# /etc/systemd/system/crond.service
[Unit]
Description=Command Scheduler
After=auditd.service systemd-user-sessions.service time-sync.target
#OnFailure=crond-notify-email@%i.service
[Service]
EnvironmentFile=/etc/sysconfig/crond
ExecStart=/usr/sbin/crond -n $CRONDARGS
ExecReload=/bin/kill -HUP $MAINPID
KillMode=process
Restart=on-failure
RestartSec=10s
ExecStartPost=/bin/sh -c "/etc/systemd/system/test.sh"
[Install]
WantedBy=multi-user.target
My "test.sh" is very simple:
#!/bin/sh
echo "CROND is restarting" | /usr/bin/mailx -s "crond failure
notification" myemailaddress at mydomain
As you can see, I have added lines "Restart=on-failure",
"RestartSec=10s" and " ExecStartPost=/bin/sh -c
"/etc/systemd/system/test.sh"" to the original crond.service
daemon file. Then, I run "systemctl daemon-reconfigure" and, from one
console, I run "kill -9 `pidof cron`" for restarting crond daemon.
After it, I receive an email... But now, if I run again ""kill -9
`pidof cron`", I don't receive any mail... I have notice that if I run
"systemctl daemon-reload" and then kill crond process, mail is sent
perfectly... but if I don't run "systemctl daemon-reload", mail is
sent ONLY first time...
Why?
I guess it is probably the way systemd works - subsequent 'crashes' (or
kills in you case) probably only restart the main process (ExecStart) and
don't re-run the ExecStartPost bits ...
I notice you have a commented out an 'OnFailure' line in the unit file -
so I guess you've been experimenting with that - which I believe is actually
the way to do what you need ?
James Pearson