I have a Linksys 3200ACM and just installed the latest snapshot (Last one was just over a month ago).
On this release I have noticed that running collectd to monitor various things, It maxes out an entire CPU Core. I thought it may be just one check, but all of them are running hot (Enabling them all puts the CPU at 100%), just putting one on you can see the load go up.
The strange thing is that looks like IO more than CPU when you run top. ilde% drops and the io% increases drastically. enabling just one check no matter what it is, maxes out the IO on one core. disabling all checks (and therefore no checks running), IO drops to 0%
Has anyone else seen this or if this is a known issue?
I have subsequently just stopped collectd in the meantime.
These are the plugins. It does not matter which one I use. Even individually the make IO run at 100%:
chrony.so
dns.so
interface.so
load.so
network.so
rrdtool.so
tcpconns.so
cpu.so
exec.so
iwinfo.so
memory.so
ping.so
sensors.so
If you have a look at the image below. this is what happens. The first 5 minutes I had only 1 running. When I added all the rest in it just spikes and then never comes down:
I was writing to a USB stick on the USB Port. I moved the RRD files to be on /tmp and I do not get any IO issues, so either there is an issue with the USB Stick or the USB driver is not working as it should
So I put in a new USB stick, seems to be running a lot better now, though io is still spiking, the load is not going up to about 3. sitting about 0.4, which is still higher than it used to be, so potentially there has been a change to the USB driver in the latest snapshots ... I will leave it to run like this for the time being
