Bug #4867
closed
Potential Memory Leak
0%
Description
I wonder if someone could help me suss out why TVHeadEnd uses an ever expanding amount of memory until it crashes.
First off, I should say I'm running linuxserver.io's docker container (with TVHeadEnd v4.3-943~g9b85e7b7c) as that could be relevant.
I run this excellent docker container on a Synology DS918+ (w/ 4GB Ram, upping this to 16GB tonight) but for the first time on Saturday, it crashed as it wanted more memory than my box had. I use Sat>IP as the source of the input signal (in cast that's relevant).
Here's what happens.
The box will start recording and the memory usage goes up by approximately 1.5MB every 2 seconds.
If the memory requirement goes beyond that available it crashes (OOM).
I'm really not sure why it's needing more memory for what should be a straightforward "receive over ip", "transfer to disk" manoeuvre. The CPU is always less than 10%, so I can only think caching. I don't think the disk i/o is a limiting factor as it's 4 x 3TB disks (Raid5) and in a "dd if=/dev/zero of=/tmp/test bs=1G" test, it completed it in just over a second.
I have played with the caching settings on TVHeadEnd under "Configuration, Recording, Recording Profile" and I'm currently on "System", but have also tried "Sync and Don't Keep", and "Don't Keep" to no avail.
Timeshift is currently enabled, but with this disabled it still displays the same symptoms.
The logs show NOTHING abnormal whatsoever and ZERO error messages related to this directly afaik.
Any ideas on tools I can run to work out why this is happening.
Thanks
:-)
dp
PS - I've posted a similar message in linuxserver.io's forums in case it's a Docker Image issue
Updated by 
Updated by