cdcron

cdcron is cron service which can also export collectd metrics.

Because cdcron uses a simple and robust time keeping algorithm, it is not suited for thousands of jobs, but in exchange is able to detect time keeping anomalies which are exported as a metric.

How cdcron handles some edge cases:

• If a job is overdue, cdcron logs it, but does not run it.
• If time jumps forward more than 30 seconds, cdcron may miss jobs but attempts to log them and export time anomaly metrics.
• If time jumps backwards more than 30 seconds, cdcron may run jobs multiple times, but attempts to log them and export time anomaly metrics.

Example

/etc/cdcron:

# All fields are mandatory
#         ┌───────────── minute (0 - 59)
#         │ ┌───────────── hour (0 - 23)
#         │ │ ┌───────────── day of the month (1 - 31)
#         │ │ │ ┌───────────── month (1 - 12, jan-dec)
#         │ │ │ │ ┌───────────── day of the week (0 - 6, mon-sun) 
#         │ │ │ │ │
#         │ │ │ │ │
job-label 0 * * * * echo 'An hour has passed'
# Repeat and range syntax
job2 */10 * * * * echo 'Every 10 minutes'
job3 0-5  * * * * echo 'First 5 minutes of each hour'

Run collectd:

$ collectd -C ./example/collectd.conf -f 

Run cdcron:

$ cdcron -metrics-mode unencrypted -cron-tab ./example/example.crontab

Example of exported metrics

The table:

job1 0/2 * * * * echo hello
job2 1/2 * * * * sleep 300

produces the following exported metrics:

host.cdcron.gauge-job1-duration 0.010488727 1629722407
host.cdcron.counter-job1-success 1 1629722407
host.cdcron.gauge-job1-utime 0.003877 1629722407
host.cdcron.counter-job2-failure 0 1629722407
host.cdcron.counter-job2-success 0 1629722407
host.cdcron.gauge-job2-duration 0 1629722407
host.cdcron.gauge-job2-maxrss-bytes 0 1629722407
...