Since version 1.3 kaffeine has a memory leak. Version 1.3.1 hasn't fixed this issue. I use kaffeine mainly for watching live TV with my two DVB-C TV cards. As soon as I let kaffeine idle the memory usage rises quickly. While watching live TV the memory usage also rises but at a much slower pace. The reason is probably the DVB EPG grabber, see attached kaffeine log. To this bug report attached is also a excerpt of my system log where kaffeine is killed because its memory usage grows to 10GiB-15GiB. When a out of memory exception happens a data loss is very probable. A few times my window manager was unresponsive and had to be killed too and with it all running programs. Reproducible: Always Steps to Reproduce: 1. Start kaffeine (memory usage: 34.1 MiB ) 2. Start live TV playback of a channel. (memory usage: 188.2MiB ) 3. Stop live TV playback (memory usage: 48.6MiB ) 4. See the memory usage of kaffeine rise about 1 MiB every second. Actual Results: Memory usage rises quickly to 1 GiB after 15 Minutes of idle time. Expected Results: Memory usage stays more or less predictable depending on the usage scenario.
Created attachment 95524 [details] Kaffeine system log
Created attachment 95525 [details] Kaffeine log of a 15 minutes test run with 1 GiB memory usage.
I am seeing a similar behaviour - when DVB-T playback runs memory consumption seems reasonably stable. As soon as the it is paused or stopped memory usage grows at about 1/2MByte extra per second. Version 1.3-git KDE platform 4.14.10 OpenSuse LEAP 42.1 64bit
I'm doing some tests here with Kaffeine 2.0.1, but I'm not seeing any memory increase. Could you please re-test with Kaffeine 2.0.1?
This seems to be fixed with 2.01 to 2.03. I have only encountered a small increase in memory usage of about 6 MiB by changing channels which are on different transponders. This way I easily managed to increase memory usage from 48 MiB to 108 MiB and the memory isn't free by stopping playback.
(In reply to lev.cohan from comment #5) > This seems to be fixed with 2.01 to 2.03. > > I have only encountered a small increase in memory usage of about 6 MiB by > changing channels which are on different transponders. This way I easily > managed to increase memory usage from 48 MiB to 108 MiB and the memory > isn't free by stopping playback. An increase as new EPG data is feed is expected. So, it seems that this issue was fixed. So, I'm closing the bug.