thestraycat Posted June 16, 2023 Share Posted June 16, 2023 As observed via top. Seems excessive to me... Can anyone confirm thats a normal usage pattern? Spec: CPU 4c/8t - 32GB ECC. Link to comment
ionred Posted June 17, 2023 Share Posted June 17, 2023 I'm in a similar boat--my numbers look very close to yours-- but I also have an open post from Monday that I haven't heard back on yet for what I'd define as excessive CPU usage for an application that should be using almost no CPU most of the time (in my opinion). Link to comment
thestraycat Posted June 18, 2023 Author Share Posted June 18, 2023 (edited) Yup same here.....I'd observed a few changes tbh since updating to the last 2 stable releases. - My idle cpu stats are much more chaotic than they have been on older unraid versions. (I usually idle between 8%-16% cpu now i idle anywhere between 15-50% which has tripled to what it would usually be a few versions back. - After boot my CPU sits at 100% on all cores for around 5 minutes before settling down which is strange behaviour seems to be when loading docker. Didnt do this on previous unraid versions. - Unraid-API seems to always be one of my biggest mem hogs which i hadn't really noticed going back a few versions but it's quite considerable. I've just rebooted the server and unraid-api is straight back up to 10.9g from a cold boot. Maybe the memory handling has changed in a recent version and what im seeing is normal, but with the process being called 'Unraid-API' it dosn't feel like it should be so high. However i have i'm unaware of what this process is doing. I was hoping to not have to go through the painstaking process of inspecting every container and plugin and thought i'd post to see if anyone was experiencing similar... Edited June 18, 2023 by thestraycat adding Link to comment
elibosley Posted June 19, 2023 Share Posted June 19, 2023 Reserved usage is not the actual program usage. See the 0.7% mem on the graph above. Reserved just means that linux has set aside some amount of memory that the program "could" use, but is not using right now. That being said, I will look into the cause of the high reservation, my guess is that it is due to the fact that the API talks to many underlying layers and those bindings may just tell linux that they need a lot of memory, even though the program is not using that much. Link to comment
thestraycat Posted June 30, 2023 Author Share Posted June 30, 2023 Agreed. However it does look less than optimal. Link to comment
Recommended Posts