Do your crons ever come a-cropper?

Közzétéve
2011-04-01 14:42
Written by
Eileen - member of the CiviCRM community and Core Team member - about the Core Team

Rumour has it that certain, ahem, consultants have been known to, on occasion, miss setting up the occasional CiviCRM cron, or not notice when the cron stops working on upgrade. Such consultants could wind up with egg on their face over crons. Somewhat less culpable are the CiviCRM newbies who didn't know their event waitlists wouldn't work or their memberships wouldn't roll over without the right cron.

 

I have noticed that amongst the current suite of MIH initiatives is the initiative to introduce a consolidated cron. So, my first suggestion is that any consultant or CiviCRM administrator who has ever had to talk fast as to why they didn't notice a cron wasn't working should immediately donate $50 to this initiative ... under a pseudonym.

 

However, I do note that the target isn't very ambitious so I'm wondering if it's a first step or a full answer to the egg-on-face syndrome. Here is what I would like to see:

 

1) an interface that shows when each cron last started & completed (in red if it didn't complete!)

2) ability to set minimum & maximum frequency for each job (if it runs less often than the minimum some form of notification would take place. It would not run more often - at least in drupal where I believe the job scheduler module handles this).

3) for drupal installs the cron would be automatically picked up by the drupal cron (scheduled per above)

4) Some notification options for if it doesn't complete - email to x, log in watchdog.

 

Assuming the current initiative falls short of this & other people feel like it would be a good idea it would be nice to try to push past the first target when it is reached to a really complete solution.

Filed under

Comments

Ken West (nem ellenőrzött)
2011-04-03 - 19:08

At the moment the various cron jobs produce various amounts of 'noise' (log messages, etc). If we're going to tie this together it would be good to get the jobs being made consistent in terms of what's output, and signalling events which need an administrator's attention.