Since a few hours ago, the process /usr/bin/python /opt/rockstor/bin/data-collector is running at 100% CPU and eating up all CPU cycles available to the machine.
Even after killing the process or after a reboot it restarts itself and caps out at 100%, staying there.
Sometimes the same is happening with /usr/bin/python /opt/rockstor/bin/gunicorn, but that seems to have fixed itself. Gunicorn also constantly takes up almost 1GB of RAM, which to me seems like way too much for a web server.
Does this high processor use also persist if you do not have a web page open on the Rockstor Web-UI Dashboard (where all the fancy graphs are). That page can take a lot of work to produce for both the client and the Rockstor machine, but usually it’s more on the client side actually: ie lots of java-script running in the browser. Also could you give an indication of the hardware you are seeing this high cpu usage on. Rockstor is not really that light weight as it’s (currently anyway) a full CentOS install. We currently have the minimum system requirement as follows:
But this is probably due to be updated to include a dual core processor and the 2 GB ram is also probably a bare minimum.
I’m not sure we actually have anything to ‘fix’ here actually. We have not previously had a report of this actually. Lets see what hardware you are running on and take it from there. Also if there is no Web-UI open at all, the most likely actual use case, then you should get even less cpu usage as there are processes that run to serve some live info in the Web-UI header also. But the Dashboard is definitely a heavy user of scraping system info to display it ‘live’.
RAM usage went down significantly when closing the web interface and gunicorn requires less CPU now, but the data collector is still running at 100%.
Rockstor is running on an ESXi VM with 2 cores and 2GB RAM with a 4GB swap partition. RAM usage sits around half full, and one CPU is maxed, the other almost idle.
I asked this in your other thread, but figured I’d ask here too, just in case someone has this problem as well.
Is your Rockstore VM configured for a single 2 core vCPU or 2 single core vCPUs? I’ve found that multiple single core vCPUs out perform a single multi-core vCPU.
Not really sure where the difference between the two options is. I am running Rockstor on a 8 core machine, of which 2 cores are assigned to the VM through ESXi.