On Thu, Feb 15, 2018 at 5:12 PM, billinghurst &lt;billinghurstwiki(a)gmail.com&gt; wrote: I assume these alerts are messages from the cron server about failures to jstart the job on the grid at the scheduled time? If your crons are actually starting grid jobs as expected I think that's the only way that cron would know that the job had output to standard out/standard error to email to you. You can check out this little tutorial on cron email related things <https://www.nixtutor.com/linux/sending-email-alerts-through-cron/> for some ideas. I'm not 100% sure if the "| mail" or ">/dev/null" style parts will actually work as expected in Toolforge. Our cron stuff is a bit abnormal in that the `crontab` program that you interact with on a bastion does some processing of the crontab data before saving it to a remote server (ping tools-cron-01.tools.eqiad.wmflabs at the moment). This pre-processing probably won't understand either of these unix shell constructs. Can we back up a half step and talk about the root issues? * What kinds of errors are you being alerted to and can they be fixed? * Why do you want to only notify some of the tool maintainers when things are failing? Are you mixing many different things under a single tool? Would it be easier to manage these special jobs from another tool account designated for the purpose? Bryan -- Bryan Davis Wikimedia Foundation &lt;bd808(a)wikimedia.org&gt; [[m:User:BDavis_(WMF)]] Manager, Cloud Services Boise, ID USA irc: bd808 v:415.839.6885 x6855