Cumulus Support

Hi Steve,

I think it was 3.2.8
I honestly can't remember, it was from a fresh install that was done on the Pi3 after doing the normal updates.

It was a revision 3 but I cant recall the last two numbers I'm afraid.

Thought I'd post an update incase anyone else had a similar issue.
Not had any issues with the USB station either so all seems good.

I've issues with certain version of mono and dependant apps before.

Am running

Mono JIT compiler version 4.4.1 (Nightly 4.4.1.0/4747417 Wed Jun 22 11:14:49 UTC 2016)

on Debian Jessie with CumulusMX now and works fine.

I know this is an old post, but I had this same issue last week as I did my first migration from Windows (10) CumulusMX to a RPI3 running Jessie Lite. The initial problem I was getting was random USB disconnects; I attributed that to having too many items connected to the USB bus (I guess the Pis can be a little finicky when it comes to USB power), so I removed everything on it. And another point -- do NOT use a USB extender...even an active one...as it did not fare any better. Once I moved the Pi closer to the Davis and plugged it into the station with the cable it came with (equipped with a ferrite core cable shield) I haven't had any more issues with the disconnects. I wrote a udev rule to symlink a dev name to the current ttyUSB* in case something did happen. Only problem is that CumulusMX needs to be restarted once this disconnect happens.

I noticed the other day I was getting the out of memory error as well. I didn't have much on the Pi running, although it was an old instance. I was running Mono 3.2.8.

I got a larger SD card and decided to redo the setup, but this time I included sources for installing mono 4.0.5. I stood up the new card with Jessie Lite, did some modifications (the udev mod again) connected and sofar it's been running just great since then. And my Twitter updates have returned.

Of note; i noticed in the logs (unfortunately I lost them in the resets) before the out of memory errors that I would get sharing errors on realtime.txt when I was trying to send them to my web server. I'm not sure that helps any, but it's what I noticed. Sofar, as I've said, it's running great.

It's possible that a failed tweet or ftp upload results in a memory leak, eventually leading to an out of memory condition. It's not something I ever experienced myself in all the time I was running Cumulus. A sharing violation, as you probably know, says that something else is doing something with the file at the same time that Cumulus is trying to write to it or upload it. That 'something else' could be another thread in Cumulus, if you have Cumulus configured to do something with realtime.txt other the normal automatic upload.

steve wrote:It's possible that a failed tweet or ftp upload results in a memory leak, eventually leading to an out of memory condition. It's not something I ever experienced myself in all the time I was running Cumulus. A sharing violation, as you probably know, says that something else is doing something with the file at the same time that Cumulus is trying to write to it or upload it. That 'something else' could be another thread in Cumulus, if you have Cumulus configured to do something with realtime.txt other the normal automatic upload.

None of that is going on with the realtime.txt. I was having problems with Twitter though..the standard mono 3.2.8 issue about invalid certificates. Installing 4.0.5 seems to have remedied the problem (I hope).