Problem scanning huge source (mem leak)?
Ok, the second JSON is able to get kodi down!

I've to say that I've hammered F5 in the browser until Maria reached its max_connections (was set to 20). I did set them higher (40) and voila... OOM!

Sadly no crashlog as the times before, but here's kodi.log

After kodi restartet I did stress again until system freeze. Once reacting again, I hammered first JSON, but that clearly has not that influence on memory, whereas each second JSON is eating up about 40MB free memory until finito...

What I am wondering is, in normal usage the queries aren't coming that fast, but in conjunction with using Xonfluence (I guess it uses LDP & S.W. queries different than Estuary, which doesn't use tham at all) I get crashes with something like "memcmp () from /usr/lib/libarmmem.so" or "malloc" which points to memory issues ... I've learned (guess it was Milhouse telling), that OOMs don't result in crashlogs, so there must be something coming together!?
I'm away now (illness is over and have to work again Wink. I let kodi run with Xonfluence, LDP & S.W. as before, although the logs aren't readable, when I had the crashes, as we now have debugging symbols and useable crashlogs.

So here's the second kodi.log and the journalctl of both OOMs

Edit
Did another try
20 of 1. JSONs=no memory usage
20 of 2. JSONs=freeze
Here's the memory.log
Reply


Messages In This Thread
RE: Problem scanning source - by DaveBlake - 2017-02-12, 13:00
RE: Problem scanning source - by Uatschitchun - 2017-02-12, 21:51
RE: Problem scanning huge source (mem leak)? - by Uatschitchun - 2017-03-07, 12:59
Logout Mark Read Team Forum Stats Members Help
Problem scanning huge source (mem leak)?0