@mwall, Thanks a lot for your analysis. I'm running an RPi B = 512 MB of RAM, if the issue show up again I can swap it for a RPi 3 that has 1GB of RAM and better CPU, maybe is overkill but I really like how weewx works. I will do some tests to see what happen.
On Monday, February 27, 2017 at 6:01:46 PM UTC-5, mwall wrote: > > On Monday, February 27, 2017 at 1:33:20 PM UTC-5, Francisco Puig wrote: >> >> I just sent the logs from the last 24hrs before and after the crash. >> > > francisco, > > how much memory is in the pi? > > you are generating only the Standard report. your reports take 20 to 80 > seconds to run. this is slow, but fairly normal for a rpi, and not a > problem since your archive interval is 300 seconds. > > you are publishing to PWSWeather, WeatherCloud, WeatherBug, WOW, and > AWEKAS. > > you are using a sqlite database, not mysql. > > the memory use of weewx was 149066/90779 VSZ/RSS just before it was killed. > > the system has no swap. > > so all of this is normal. > > the system probably decided to kill weewx because weewx has the biggest > total_vm (VSZ), even though weewx is not doing anything unusual. > > at 06:33:06 the kernel started having problems. just before that, apache > was reloaded and failed. > > at 06:45:26 the kernel killed weewx (pid 472) > > after the kernel killed weewx, systemd started apache again. this time it > was successful. > > my guess is that apache was eating up memory, the system tried to restart > it, that failed, then the system killed weewx. > > try disabling apache and use nginx or lighttpd instead. or, if this is a > 512M or 256M pi, get rid of the display and disable all video/graphics, and > connect using ssh. > > m > > -- You received this message because you are subscribed to the Google Groups "weewx-user" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
