We started to receive a lot of crashes. We need to have a good way to distinguish them, so that we don't simply do "Mark all read" and miss some new important crashes hidden in a heap of we-know-about-this crashes. Possible improvements I can think of:

• Distinguish dev/stg/prod environment. Please include [dev], [stg] or [prod] in the email subject.
• Include step name in the email subject, e.g step: runtask. That will help us distinguish check crashes from infrastructure crashes (git cloning, etc)
• Include the first line of the log in the email body. Currently, it's very hard to see what command was run from just reading the email, you have to click on the link. If you include the first line of the log, it will be clear, because it looks like this (at least when looking [[ https://taskotron.stg.fedoraproject.org/taskmaster/builders/x86_64/builds/13607/steps/runtask/logs/stdio | here ]], I'm not sure the blue text is reflected in the underlying file. But it should be stored somewhere?):

runtask -i f20-updates-testing-pending -t koji_tag -a x86_64 -j x86_64/13607 upgradepath.yml

• Include task name in the email subject. Getting check name directly from buildbot is hard (according to @mkrizek). But if we can access the first line of the log, we can parse it from there! So let's just take the last word of the line (upgradepath.yml) and strip .yml. And we have the task name.

The subject can then look e.g. like this:

Taskotron failure (upgradepath, dev, x86_64, step: runtask)