Need CPU limit to keep my old nas working!

[mjl]

I have an old TS-109 Pro. The problem is that with the current firmware (Version 3.3.0 build 0924T) and SABnzbd+ (0.5.0) installed then NAS is hardly uasble.

When SABnzbd is active it occupies the CPU with 92-99.9% (looking at TOP usage statistics).

In this way file sharing, etc is extermely slow!

I found that in Linux one can limit the CPU of a process using "cpulimit". This is not available in the firmware. I tried getting it through the Optware (iPKG) system but there it does not exist either.

Is there any (command line) tool to limit the CPU usage of the SABnzbd.py process? If not: definite feature request!!

[schumaku]

I don't think forcing a CPU limit does help anything - CPU is just one of several factors, other factors are memory and I/O. Why? When other processes need CPU, they will get thier part, say about 50% when two processes are fighting for CPU.

Not a Sab'crack - but I would consider to lower whatever is possible: Number of concurrent jobs, bandwidth available, ...

[mjl]

I don't think forcing a CPU limit does help anything - CPU is just one of several factors, other factors are memory and I/O. Why? When other processes need CPU, they will get thier part, say about 50% when two processes are fighting for CPU.

Not a Sab'crack - but I would consider to lower whatever is possible: Number of concurrent jobs, bandwidth available, ...

SABNZBD gets (and takes) the full 100% and thereby makes the NAS useless for other purposes. I tried to compile CPULIMIT, however GCC does not work on this nas, so that is a nogo...

If I limit the IO it still hogs the system with 100%.

Setting SABNZBD to 70% with CPU limit would be enough, file sharing only takes 30% max when active.

So if anyone out there knows how to compile (or send me the binary) of CPUlimit for a good ald 109 I'd be happy!

Going back to an older firmware is no option anymore (4k clusters...), although, i'd be happy if i could, never had these issues on the older firmware, the new firmware is just to much for this hardware.

[schumaku]

Dount there were much changes on the Kernel side from v2/early v3 to the one supporting 4k disk blocks.

A process able to use CPU pow available does not mean the CPU is no longer available for other processing - neitther on the NAS, on Linux in geneal, nor on any other utlitasking OS in my opinion. Priority might be to high...

[mjl]

Dount there were much changes on the Kernel side from v2/early v3 to the one supporting 4k disk blocks.

Indeed, I don't have data loss anymore, caching has been fixed at some point....

A process able to use CPU pow available does not mean the CPU is no longer available for other processing - neitther on the NAS, on Linux in geneal, nor on any other utlitasking OS in my opinion. Priority might be to high...

How to change the priority then, is that easy on a QNAP ?


--> found it: "renice", works in part, I can now access the drive from windows, transfer is still very slow, but that is probably due to the IO issues. Would be nice if "nice" would also be available, I could give low priority to admin related processes, giving the users a bit more room for connecting.

[fantomas]

Renice does exactly the same as nice does. the only difference is that nice executes a processwith different priority, while renice changes priority of already running process.

Different thing youmight want is ionice, which maintains i/o priority, while (re)nice only affect CPU priority.
I don't know if ionice is available and if QNAP uses "cfq" IO scheduler that is needed for ionice to work.