2 ms·
Thanks for the heads up. We're rebuilding some infrastructure now. We would have defaulted to building on 14.04, but this gives us reason to switch to 15.10.
by bryanlarsen 11y ago
Thanks for the heads up. We're rebuilding some infrastructure now. We would have defaulted to building on 14.04, but this gives us reason to switch to 15.10. That's a change I've been wanting to make for a long time anyways, but I was waiting for 16.04 LTS. Yesterday a major portion of our site was down for a few minutes because an `service nginx reload` silently failed because upstart had lost track of the process. I assume that sort of thing wouldn't happen with systemd because it uses cgroups.
- Ambroos 11y agoWe've switched to Debian Jessie over 14.04 a few months back for systemd and more up-to-date dependencies too. I'm not much of a sysadmin guy (developer-turned-Ansibleman), but our experience with systemd has been amazing so far for managing our sockets, PHP pools and webservers. Much smoother than anything we've had before. We've prepared for this problem too. Our SSL termination is handled on a completely independent server that basically just forwards requests to internal application servers, and it's completely stateless. This way we can just set up a new one on a less-stable distribution (we're experimenting with Debian Stretch) and switch once we find it stable enough to do just that. It means we can reap the benefits of possibly unstable HTTP/2 without affecting our stable application/storage stack.