Using zabbix to monitor output from a cron job

[luvdablus]

I am new to zabbix. I have a cronjob that runs on a redhat linux server that shows me the % of tablespace of an oracle db, but still takes into account autosizing. I want to make zabbix agent aware of this cron output and report it.
Is this possible?

[kloczek]

https://www.zabbix.com/documentation/3.4/manual/appendix/items/trapper

[bbrendon]

https://www.zabbix.com/documentation.../items/trapper

I don't see how that link is relevant...

[kloczek]

I don't see how that link is relevant...

Trust me it is, because in case of cronjob you may know when job starts (less likely but still some people may be interested when it started) and when cronjob ends, and with what kind of the status.
Because how cronjobs are working (as a bit asynchronous events) you will never know when exactly it started and finished.
To monitor async things trapper items are the proper type of the zabbix items.
In other words to monitor exact type of the cronjob you may need 1 to 3 metrics:
- status of the cronjob
and optionally:
- start time
- end time
Using above you may organise for example backups monitoring because end_time-start_time will give you how long backup have been running (to monitor is this time does not exceeds your backup window) and first metric will give you possibility to define second trigger about status of the backup.
All what you need is adding zabbix_send command into your backup script.

In you case you want to monitor only output of the cronjobs. This as well asyns metric. You need to organize just one text item which content will be delivered over trapper item as well.

[Markku]

I have a Perl script that is run daily in cron to do the MySQL database partition maintenance. In the Perl script I added quick’n dirty Zabbix trapper:

printf("Sending Zabbix trap to server\n");
my $dt = DateTime->now;
$dt->set_time_zone($curr_tz);
my $dtstr = $dt->ymd . " " . $dt->hms;
system("/usr/bin/zabbix_sender -z zabbix.server.address -s hostname.of.this.server -k mysql.partition_maintenance -o \"Run at $dtstr\"");

Then, in the Zabbix server I added that mysql.partition_maintenance item (type of character if I remember correctly), and I created a ”nodata” trigger for that item. Thus, if the maintenance script is not executed for some specific amount of time, the trigger will notify the admins. In this example the contents of the trapper item data is not relevant (because just ”nodata” is used) but you could also include some status message in the data to check that in the trigger if needed.

Markku

[kloczek]

I have a Perl script that is run daily in cron to do the MySQL database partition maintenance. In the Perl script I added quick’n dirty Zabbix trapper:

BTW patritioning.
Your way of maintaining partitions has some embedded issue.
If you DB backend will be used on master/slave and time to time yoy willl be moving master DB you must always remember about move your script to new location.
It is solution which has no this issue which is using SQL event.
SQL event is SQL engine internal like cron job manager and on define it yu can specify even does it needs to be executed on master or slave DB,
As event executed on master is perfect for maintaining partitioning similar event executed on slave can be used to schedule make backup (lock DB -> flush-> create snapshot and copy it off site or even execute mysqldump -> unlock database)
Good thing about that way is everything is in database, Making database dump you can specyfy dump as well database events. Importing such dump will automatically initiate scheduling database events.
Only thing which is necessary to take care is make sure that MySQL engine event scheduler is switched ON, and this is why in my MySQL monitoring template is metric about checking
"SELECT variable_value FROM performance_schema.global_variables WHERE variable_name='event_scheduler';" and activated trigger SVC::MySQL::event_scheduler=OFF if this query des not return "ON"

mysql> show events\G;
*************************** 1. row ***************************
                  Db: zabbix
                Name: e_part_history_manage
             Definer: root@localhost
           Time zone: SYSTEM
                Type: RECURRING
          Execute at: NULL
      Interval value: 1
      Interval field: DAY
              Starts: 2017-12-14 00:02:00
                Ends: NULL
              Status: ENABLED
          Originator: 0
character_set_client: utf8
collation_connection: utf8_general_ci
  Database Collation: utf8_bin
*************************** 2. row ***************************
                  Db: zabbix
                Name: e_part_trends_manage
             Definer: root@localhost
           Time zone: SYSTEM
                Type: RECURRING
          Execute at: NULL
      Interval value: 2
      Interval field: WEEK
              Starts: 2017-12-14 00:02:00
                Ends: NULL
              Status: ENABLED
          Originator: 0
character_set_client: utf8
collation_connection: utf8_general_ci
  Database Collation: utf8_bin
2 rows in set (0.00 sec)


mysql> SHOW CREATE EVENT e_part_history_manage\G; SHOW CREATE EVENT e_part_trends_manage\G;
*************************** 1. row ***************************
               Event: e_part_history_manage
            sql_mode: ONLY_FULL_GROUP_BY,STRICT_TRANS_TABLES,NO_ZERO_IN_DATE,NO_ZERO_DATE,ERROR_FOR_DIVISION_BY_ZERO,NO_ENGINE_SUBSTITUTION
           time_zone: SYSTEM
        Create Event: CREATE DEFINER=`root`@`localhost` EVENT `e_part_history_manage` ON SCHEDULE EVERY 1 DAY STARTS '2017-12-14 00:02:00' ON COMPLETION PRESERVE ENABLE COMMENT 'Zabbix history tables partitioning.' DO BEGIN
        CALL partition_maintenance('zabbix', 'history',        28,    24,    7);
        CALL partition_maintenance('zabbix', 'history_log',    28,    24,    7);
        CALL partition_maintenance('zabbix', 'history_str',    28,    24,    7);
        CALL partition_maintenance('zabbix', 'history_text',    28,    24,    7);
        CALL partition_maintenance('zabbix', 'history_uint',    28,    24,    7);
    END
character_set_client: utf8
collation_connection: utf8_general_ci
  Database Collation: utf8_bin
1 row in set (0.00 sec)

ERROR:
No query specified

*************************** 1. row ***************************
               Event: e_part_trends_manage
            sql_mode: ONLY_FULL_GROUP_BY,STRICT_TRANS_TABLES,NO_ZERO_IN_DATE,NO_ZERO_DATE,ERROR_FOR_DIVISION_BY_ZERO,NO_ENGINE_SUBSTITUTION
           time_zone: SYSTEM
        Create Event: CREATE DEFINER=`root`@`localhost` EVENT `e_part_trends_manage` ON SCHEDULE EVERY 2 WEEK STARTS '2017-12-14 00:02:00' ON COMPLETION PRESERVE ENABLE COMMENT 'Zabbix trends tables partitioning.' DO BEGIN
        CALL partition_maintenance('zabbix', 'trends',        730,    336,    2);
        CALL partition_maintenance('zabbix', 'trends_uint',    730,    336,    2);
    END
character_set_client: utf8
collation_connection: utf8_general_ci
  Database Collation: utf8_bin
1 row in set (0.00 sec)

https://dev.mysql.com/doc/refman/5.7...ow-events.html

[Markku]

Hi, thanks for the note, I have recognized the situation and will look into it. In the wiki there were good instructions for that as well I believe.

Markku

[kloczek]

Hi, thanks for the note, I have recognized the situation and will look into it. In the wiki there were good instructions for that as well I believe.

Last small detail about partitioning monitoring.
Instead monitoring status of the execution partitioning procedure which needs to be done using trapper item(s) it is easier to check does exist next day/week/month partition on exact table and than on top of this create trigger to warn that something is wrong with partitioning.