I remember once haranguing a group of Java developers to use the log4j's RollingFileAppender. No matter what arguments I used and despite many nods and promises, on live systems they just kept on using the default FileAppender like there's no tomorrow, despite it being a relatively trivial config change to switch to the RollingFileAppender.

Only the sysadmins "got it" because they were the ones that got to feel the pain of cleaning up a NAS filled up by log files.

To my mind the whole point of DevOps is to make the developers "get it" by making them feel the pain when they get things wrong. Developers make decisions about the system every day, so it pays for them to understand how it is maintained and operated.

My impression is that devops instead has turned into a means for getting sysadmins out of the way so developers can use the latest shiny in production...

Out of interest, why wouldn't you configure logrotate on the machine itself, rather than making this an application concern?

Presumably logrotate was already set up for the variety of other logs on the system?

I was curious too, though it seems log4j just keeps writing to the old log (it doesn't re-open the file on SIGHUP), and copytruncate is not completely safe from data loss.

Did they own the systems their applications were deployed to? Because that is the essence of full stack development. Otherwise, config management is required as boundary context between developers and operations.

This only proves your Java developers were stupid and arrogant; using RollingFileAppender or something even smarter would be the norm in any sufficiently functional organization, regardless of whether developers anticipate the need or sysadmins need to ask.

I love this analogy. SysAdmins need to have an special mindset that is diametrically opposite than these of the developers.

Developers want features and shiny things and following new trends and unicorns.

SysAdmins want 99.999%, and dependability and knowing that all the backups are running on time and writings contingency plans detailing how many minutes it will take to put a new infrastructure online in the event a fire occurring in the main data-center while the backup data-center is flooded because of a typhoon.

He does say in the article that good devs can make passable DBA's, deployment engineers, etc.

Of course they are not going to be experts at it, because their field is development.

the same way, guitarists usually hit the right tones on a bass :D