I've not done anything completely retarded, but a few months ago I accidentally moved some mailboxes onto the wrong server and nearly caused the log drive to run out of space.
Luckily my colleague was on the ball and caught it before it did, as it was occuring over a weekend. Not made that particular mistake since.
Also when we fail over Exchange clusters we generally run the Stop-ClusteredMailboxServer command then just restart the node, all the resources move over, then we run Start-ClusteredMailboxServer. In the case of one of our clusters I forgot it was a CCR cluster and not a SCC cluster so this did not work so well as it was missing logs it needed. We only have one CCR cluster so in my defence it was just me being stupid.
In the end one of the databases would not mount and I had to run Restore-StorageGroupCopy to force it to mount with one or two logs missing. Learnt my lesson there too
Considering I log onto all servers all day long I have never shut a machine down instead of logging off. My colleague did though and I ripped him for a bit (in good spirits ofcourse, could happen to anyone!).