Fix trigger in Zabbix 4.0 LTS, add support to systemd 219 (CentOS 7)

README.md

@@ -24,15 +24,16 @@ Installation
 ------------
 
 * Server
- * Import template `Template_App_systemd_Services_v4.xml` file if you have Zabbix 4
- * Import template `Template_App_systemd_Services_v5.xml` file if you have Zabbix 5
+ * Import template `Template_systemd_Services_v4.xml` file if you have Zabbix 4
+ * Import template `Template_systemd_Services_v5.xml` file if you have Zabbix 5
  * Link template to host
 * Agent
  * Place the following files inside `/etc/zabbix/`:
  * `service_discovery_blacklist` or `service_discovery_whitelist`
  * Place the following file inside `/usr/local/bin/`:
  * `zbx_service_restart_check.sh`
  * `zbx_service_discovery.sh`
+ * If you have an old systemd version (before 230), adopt the file `_old` version instead, removing its `_old` suffix
  * Set executable permissions on both scripts:
  * `chmod +x /usr/local/bin/zbx*.sh`
  * If running SELinux run restorecon on the two scripts in `/usr/local/bin` e.g.:
@@ -82,6 +83,13 @@ zabbix_agentd -t "systemd.service.status[sshd]"
 zabbix_agentd -t "systemd.service.restart[sshd]"
 ```
 
+The `status[SERVICENAME]` will return:
+
+* `[t|1]` if that service is running
+* `[t|0]` if that service is NOT running
+
+The `restart[SERVICENAME]` will return the most recent restart date.
+
 License
 -------
 

Template_App_systemd_Services_v4.xml → Template_systemd_Services_v4.xml

@@ -195,7 +195,7 @@
  <tags/>
  </trigger_prototype>
  <trigger_prototype>
- <expression>{SystemD service monitoring template:systemd.service.status[{#SERVICE}].last()}&lt;&gt;0</expression>
+ <expression>{SystemD service monitoring template:systemd.service.status[{#SERVICE}].last()}=0</expression>
  <recovery_mode>0</recovery_mode>
  <recovery_expression/>
  <name>{#SERVICE} service not running</name>

usr/local/bin/zbx_service_restart_check.sh

@@ -8,6 +8,7 @@ uptime=$(date +%s -d "$(systemctl show --value -p ActiveEnterTimestamp sysinit.t
 if [[ $(( now - uptime)) -lt 180 ]]; then
  echo 0
 else
+ # note that the next line requires systemd version 230 because of "--value"
  service_start=$(systemctl show --value -p ActiveEnterTimestamp "$service")
  service_start_as_epoch=$(date -d "$service_start" +%s)
 

usr/local/bin/zbx_service_restart_check_old.sh

@@ -0,0 +1,31 @@
+#!/bin/bash
+#
+# Version supporting old systemd versions (before 230)
+# For example it supports systemd 219
+#
+
+service="$1"
+now=$(date +%s)
+
+uptime_value_raw=$(systemctl show -p ActiveEnterTimestamp sysinit.target)
+
+# parse the value that is something like "ActiveEnterTimestamp=SOMETHING"
+# https://unix.stackexchange.com/q/728424/85666
+uptime_value=$(echo "$uptime_value_raw" | cut -d"=" -f2)
+
+# Don't alert if the server has just been restarted
+uptime=$(date +%s -d "$uptime_value")
+if [[ $(( now - uptime)) -lt 180 ]]; then
+ echo 0
+else
+ # avoid to use the "--value" since it was introduced only in systemd version 230
+ service_start_raw=$(systemctl show -p ActiveEnterTimestamp "$service")
+
+ # parse the value that is something like "ActiveEnterTimestamp=SOMETHING"
+ # https://unix.stackexchange.com/q/728424/85666
+ service_start=$(echo "$service_start_raw" | cut -d"=" -f2)
+
+ service_start_as_epoch=$(date -d "$service_start" +%s)
+
+ [[ $(( now - service_start_as_epoch )) -lt 180 ]] && echo 1 || echo 0
+fi