I think by now it's cleat that the clock problems aren't problems in the RTC, but simply that somehow the system hangs (for 5mins at a time) when it gets in too-deep sleep, leading to the system clock not advancing for 5minutes (among other problems, but since this is only when the system deep-sleeps, it is usually at times when nothing else happens, so the most noticeable effect is the clock losing sync).
The problem has been known for a while, so I'm wondering if there's any progress is tracking it down and fixing it (as opposed to working around it by disabling the deep sleep states).
Are the newer 2GHz/2GB boxes affected as well?
Any news regarding the lost interrupts?
Re: Any news regarding the lost interrupts?
There is no complete solution to the problem but there are more subtle workarounds listed at http://fit-pc2.com/forum/viewtopic.php?f=57&t=1595
Re: Any news regarding the lost interrupts?
The workarounds may fix some of the problems, but they don't prevent the system from freezing 5minutes at a time (presumably when going into deep sleep), which makes things like the watchdog unusable since the 255s<5min.irads wrote:There is no complete solution to the problem but there are more subtle workarounds listed at http://fit-pc2.com/forum/viewtopic.php?f=57&t=1595
And as far as my experience goes, ntpd is not a workaround, since ntpd will just give up. Maybe OpenNTPd handles it better, but at least ntpd under Debian was of no use when I tried it.