A Brief Retrospective on the Sprite Network Operating System

bediger4000•2 days ago

Walnut Creek sold a Sprite CD-ROM back in the day. I got Sprite to boot on a Sun SPARCStation IPC. It was kind of unstable. I really wanted to try process migration and the log structured file system.

spijdar•about 7 hours ago

You really don't want to run Sprite on a single host, unfortunately -- the local filesystem path is buggy, and there are some major bugs in the system as preserved on the CD-ROM, e.g. vfork() is totally borked, at least for software which expects a BSD-style vfork.

With some fixes, it can be pretty stable. Here's a web server running on Sprite on an IPC: https://ember.shockfox.net/

Or, well, it is running right now. I don't leave it online 24/7, so caveat emptor to future visitors.

Qwuke•about 7 hours ago

Is there anything I can read about the fixes to make it more stable? Or just a place to check out recent developments?

spijdar•about 7 hours ago

At the moment, no. Beyond just fixing problems I've found, for the last year I've been doing a lot of haphazard development inside of Sprite itself, so "fixes to bugs in released Sprite" are intermingled with "features I've added". The bigger problem is that getting data in/out of Sprite is ... harder than you'd imagine.

I am hoping to make a big writeup "soon", since there's a lot of interesting information about Sprite that's mostly been lost to time, where Sprite is mostly remembered as having "process migration" and "LFS". Meanwhile, it's mostly forgotten that Sprite was (apparently) one of the first OSes to adopt a mostly-GNU userland, had a modern event-loop style abstraction in C for handling I/O and timers (Fs_Dispatch), had a userspace filesystem abstraction akin to FUSE (PFS), implemented a PTY-like system entirely in userspace (the Td* functions in libc), and overcommitted memory by default.

EvanAnderson•1 day ago

Process migration seems like something that should be commonplace today. It would be an amazingly handy feature.

jacobgorm•about 6 hours ago

I invented VM live migration to solve the problems with traditional process migration, and my work was inspired by both Sprite and MOSIX. I believe that VM migration is in wide use today, especially at hyperscalers. Mike Nelson, who worked on Sprite, invented VM live migration in parallel with me, but secretly within VMware.

toast0•about 8 hours ago

I feel like if you have a single system image, process migration is easy (or at least tractible), but without that, it's going to be pretty challenging. Live migration of VMs across hypervisors is more approachable than migrating a process between two individual systems... You've got all that I/O that needs to be proxied, etc.

Single system image across multiple nodes doesn't seem to be getting much mindshare either.

convolvatron•about 8 hours ago

I did a research SSI in the early 2000s. After getting it largely to work, we kind of put it aside. Why would we aspire to using 'ps' and 'ls' and 'mount' and such to manage our large scale systems? surely there was a better model.

20 years later people are still sshing into nodes to fuss around, and treating cluster nodes like pets. it still could be done pretty easily, I guess its not really in anyone's interest to invest the 2 years it would take. there is a lot more money to be made nibbling around the edges of the problem.

A Brief Retrospective on the Sprite Network Operating System

⚡ Community Insights

Discussion (11 Comments)Read Original on HackerNews