|
|
|
|
|
by LeonenTheDK
958 days ago
|
|
I'm curious if you could talk about any of the consequences at all. There's a very similar situation unfolding at my company currently. New project running on FreeBSD because that's what the admin likes, and he takes every opportunity to bash Linux that he can. I'd like to prepare for potential pitfalls best I can. We've already had issues around compatibility with third party software and OS upgrades (which surprised me in particular, likely a mistake on the admin's part but I'm not privy to the details). FreeBSD seems like a wonderful OS for purpose-built hardware (eg pfsense, PlaystationOS, Netflix's distribution servers), but I'd much prefer to be on something common for the general compute. I don't even want to know what hiring another/replacement admin will be like in the future (we're entirely on-prem in a city that's away from a major population centre, niche enough as it is without adding FreeBSD on top). |
|
One major consequence is that your software and setup scripts are likely to continue to work over many major releases. Basic system tools don't often get replaced, so you don't need to worry about migrating to a new tool to configure interfaces, or a new firewall (although you do have three firewalls to choose from, they all have been there for a long time) or a drastically new init system etc.
Otoh, there are many projects that are made for Linux only. Sometimes intentionally, sometimes unintentionally. Lots of assumptions creep in, people write Makefiles that need GNU make, and call make recursively with 'make' instead of $(MAKE), but FreeBSD has BSD make as make, and GNU make, if you install it, as gmake. FreeBSD sets up cc to be llvm's clang, and I don't think that's common in Linux. Many projects compile with either, but some don't. Specialized linking is tricky. Compiler and linker changes are probably an exception to my rule earlier about not replacing tools, but there's a case to be made that clang is at least as good as gcc, and the license is a better fit; there's no requirement that the base be all BSD licensed, but it's nice to move in that direction where possible.
System monitoring is different, but capabilities are roughly similar. FreeBSD has some nice things, and Linux has some nice things, I've got my preferences, but whatever.
I don't think it's hard to take a reasonable systems aware person and sit them down with a FreeBSD book and a FreeBSD system and get acceptable results. Yes, mastery takes time, but consultancies are available if you need mastery right now, and don't have time to develop it in house.