Add this to your 2038 bug list, a talk by Thorsten Kukuk of SUSE at All Systems Go 2023!

2023-09-13, 16:50–17:15 (UTC), Main Hall

https://cfp.all-systems-go.io/all-systems-go-2023/speaker/PU9L3K/

https://cfp.all-systems-go.io/all-systems-go-2023/talk/3Z7XEE/

"The utmp implementation of glibc uses on quite some 64bit architectures a 32bit time variable, which leads to an overflow on 03:14:07 UTC on 19 January 2038. This talk will explain the current work on replacing utmp with logind."

[ #2038 #y2k38 #y2038 #systemd #utmp ]

cc @thilo

The scary thing about the #Y2038 problem is that the only systems likely to fail are the embedded sort of systems; systems whose failure would be incredibly destructive.

Anyway, these are the things I think about on a Sunday evening. How was your weekend? 😂

#unix #EmbeddedSystems

@dalias @thesamesam The OS infrastructure work (eg, in glibc) is important pre-req work, but it is only a start. By the same metric, #IPv6 would have been "done" in the mid-2000s when most OSes added support. It is all the application software, protocols, databases, embedded systems, and other long tail of stuff that both need changes and in some cases roll-outs for #Y2038. At some point soon we will critically need ongoing testing and qualification/certification for readiness.

@dbelson @hanscath I also share the concern that the relative success of #Y2K mitigation efforts will lull us into catastrophe, and/or that we'll all spend our last years before retirement as consultants focused on the #Y2038 problem, with our kids doing the same sort of tactical work many of us did for Y2K.

Some academic studies on the scope might be really interesting, e.g. testing out samplings of embedded systems and open source, looking at how many protocols and DBs use 32-bit values, etc.

Next Thursday we'll be only 15 years out from #Y2038, the date when a 32-bit signed time_t rolls over. It is already past time to start planning for it. Many of the cars and devices sold today will hopefully still be functioning in 15 years, so should be certified now that they won't break then. Also, timestamps in the future (eg, expiration dates) will increasingly cross the threshold as we get closer.

I wrote a blog in 2019 which still applies today: https://www.akamai.com/blog/performance/preparing-for-y2038-already

@dave_andersen Given how long the expiration date usually is on gefilte fish, it must have been purchased in the 90s... (I'm now wondering what long shelf life products will start having #y2038 issues first now that we're 15 years out.) I usually make my own with tilapia, because it is a good vehicle for horse radish.

Is there an organization working to fix the #Year2038 problem across all of computing? I imagine something like "The International 2038 Task Force" or something like that.

More info: https://en.wikipedia.org/wiki/Year_2038_problem

#y2038 #y2k #y2k38 #epochalypse

@kensanata Oh boy, you're forgetting the 2038-01-19T03:14:07 UNIX epoch situation we have to survive before. I assume this is more dangerous than the #Y2K situation since we can live without #Windows server but when non-prepared UNIX and Linux fails ... 😱
https://en.wikipedia.org/wiki/Year_2038_problem
#Y2038 #Epochalypse