Johnny Billquist <bqt%update.uu.se@localhost> writes: > On 2020-10-10 15:43, Greg Troxel wrote: >> The real fix is to add an RTC. While I know the PDP-11 didn't have one, >> and the operator would run date on boot, it's been assumed by UNIX for a >> long time that there is a clock. I have a clock addon for the RPI that >> I haven't gotten around to installing; in theory it slips over the GPIO >> pins and will still fit in the case. > > Careful with that axe, Eugene. > The 11/9x machines do have a battery backed clock. > Also 2.11BSD is often talking ntp, so no operator setting the clock > manually at boot time. OK - but I meant the 11/40 in 1977. And my 11/23 at home did have one because my father and I built a qbus card with an RTC chip! > That dns starts failing if you don't have a correct clock seems to be > a serious brokenness. Plenty of embedded devices that might not have a > battery backed clock... It's not that "dns fails". It's that you have to decide if you want to validate signatures, or you don't. If you are configured to validate signatures (with validity intervals on keys), then you need time. So embedded things without a clock need to refrain from signature validation in their config. I have several ESP8266 devices that do perform DNS resolution just fine and they have no idea what time/year it is. The problem here is that the RPI is being treated like a Real Computer and it isn't quite.
Attachment:
signature.asc
Description: PGP signature