High CPU usage on install and (kernel) update

VitalPBX Community Support General Discussion High CPU usage on install and (kernel) update

  • Post
    erik
    Participant

    I have recently installed VitalPBX on a Supermicro X7SPA-D525-HF. This is a passively cooled system. At the end of the install, which took a long tim, it started giving me temperature warnings. Luckily it finished before frying the CPU.

    Then I did a yum update to bring the system to the latest version (2.3.8-3). This also installed a new kernel and during update-initramfs I was getting the same warnings again. It too took a while, so I am guessing during the install it may have been update-initramfs as well.

    I am running Debian systems on other systems with the same hardware and never noticed anything like this. Yes, update-initramfs takes a while on them too, but not as long and it certainly does not heat the CPU as much.

    0
Viewing 10 replies - 1 through 10 (of 10 total)
  • Replies
    Up
    0
    Down

    Do you data or tenants created on the system?

    0
    erik
    Participant
    Up
    0
    Down

    @ing-joserivera26

    I am not sure I understand what you’re asking. This is a clean install from the ISO, followed by a yum update. Only installed what the ISO installer installed.

    Not a multi-tennant system, if that’s what you’re asking.

    There are 9 extensions and 2 trunks, but obviously those don’t affect the OS. I am not too familiar with CentOS. All the other systems I’ve built have been Debian pr Debian based.

    I don’t recall if the same effect was present with Elastix.

    Maybe it’s more of a CentOS question than a VitalPBX?

    0
    Up
    0
    Down

    Did you made a full update? From what version did you made the update?

    0
    erik
    Participant
    Up
    0
    Down

    @ing-joserivera26

    This is a clean install from the latest 2.3.8 ISO offered by the ayera mirror, dated 4 oktober.

    Then I did a “yum update” to bring the system to the latest version (2.3.8-3).

    Recap: Full update from 2.3.8 to 2.3.8-3

    0
    Up
    0
    Down

    Got it, I can be possible that be something related to Centos 7, or the virtulization your using.

    0
    erik
    Participant
    Up
    0
    Down

    @ing-joserivera26

    That must be CentOS then. No virtualisation. It’s running on a Supermicro X7SPA-D525-HF, bare metal. This is a passively cooled system with an Atom processor.

    0
    mo10
    Moderator
    none
    Up
    0
    Down
    Posted by: @erik

    @ing-joserivera26

    That must be CentOS then. No virtualisation. It’s running on a Supermicro X7SPA-D525-HF, bare metal. This is a passively cooled system with an Atom processor.

    There is clearly something wrong with your cooling!
    Even the mainboard is passive cooled you need at least a good airflow with fans inside the enclosure. 

    See here and other google searches.
    https://forums.servethehome.com/index.php?threads/supermicro-x7spe-hf-d525-cooling-suggestions.13972/

    Sure the Installation and Update auf VitalPBX uses the CPU. This is normal. Even after hours of 100% CPU Load your system should not get too hot.

    0
    erik
    Participant
    Up
    0
    Down

    @mo10

    Thanks.

    The entire system is passively cooled and has been for 10+ years. There are no fans inside the enclosure. There is normal airflow.

    Yes, it is to be expected that the installation and update of VitalPBX uses more CPU than normal operation, but that does not cause a noticeable rise in temperature. It is specifically during update-initramfs that the temperature went way up. Next time, I will be ready to collect more data.

    Also, I never said it got too hot. It just got hot enough to raise temperature warnings on the BMC. Could it have sustained this level of operation for hours, maybe?

    The point was that it got hot at all. I have been running Elastix on this system for many years and never noticed anything like this when installing or updating.

    I don’t think Elastix ever got beyond Centos5, so maybe this is normal behaviour for Centos7. I have no data to compare and was wondering if anyone else has.

    0
    mo10
    Moderator
    none
    Up
    0
    Down

    Install a clean debian and install:
    stress or stress-ng (CPU Stresstesting)
    lmsensors (Temperatures)

    Then watch the Temps and try again with CentOS.

    example for stresstesting:
    stress -c 4
    or
    stress-ng -cpu 4

    Share your temperature-results

    Keep in mind that a Intel Atom D525 is not really a fast CPU (very slow). This is why the VitalPBX installation takes so long.

    0
    erik
    Participant
    Up
    0
    Down

    @mo10

    Thanks, but this is a production system. No room for experiments. When a test system becomes available, I will do some more digging. Not anytime soon, sadly.

    0
Viewing 10 replies - 1 through 10 (of 10 total)
  • You must be logged in to reply to this topic.