When searching for test faillures, my workflow is as follows:

> fetchlogs -repo tools -n <some large number, usually 3000>

Followed by

> greplogs -md -l -e 'FAIL: TestName' -since=<date>

The problem is, if my -n value is too small I may silently miss test failures in my -since range. Also, without doing a bit of digging it's not easy to see how far back fetchlogs results go.

A nice solution to this problem would be to add a -since flag to fetchlogs, telling it to keep searching dashboards until it finds a build where both the repo revision date and go revision date are too old.

CC @bcmills