CloudStack
  1. CloudStack
  2. CLOUDSTACK-3415

Guest os memory is not getting dynamically adjusted incase of overcommit

    Details

    • Type: Bug Bug
    • Status: Closed
    • Priority: Critical Critical
    • Resolution: Fixed
    • Affects Version/s: 4.2.0
    • Fix Version/s: 4.2.0
    • Component/s: Management Server
    • Security Level: Public (Anyone can view this level - this is the default.)
    • Labels:
      None
    • Environment:
      master stable branch ;vmware esxi-5.1

      Description

      In case of contention existing vms memory should automatically get reduced (should be between reserved and max) so that new vms can accommodate on host.But i did not see this behavior on vmware.
      Guest OS always shows max memory even in case of contention .

      Steps to reproduce
      ---------------------------
      1-Preapre a CS setup with vmware esxi5.1
      1-Set memory overcommit to 2
      3-Create a service offering of1 (cpu=500MGZ,Memory=4096MB)
      4-Deploy vms using service offering of1 till no memory left on hypervisor
      5-run cat /proc/meminfo on each guest

      Actual
      ----------
      1-Step 5 shows each guest is having 4096MB
      2-Total guest os memory(at guest OS level) >what physically available on host

      Expected
      --------------
      1-VMs memory should dynamically get adjusted (<4096) in order to accommodate new vms on host
      2-Total guest os memory (at guest OS level) <=what physically available on host

      1. Logs_DB.rar
        1.89 MB
        prashant kumar mishra

        Activity

        Hide
        Bharat Kumar added a comment -

        The memory shown at the guest os level is not correct as the guest OS is not aware of the memory optimization at the hypervisor level. In order to verify the actual allocation vmware counters should be observed.

        In this case as mentioned total memory > actual memory available. which means overcommit is working.

        this is not a bug.

        Show
        Bharat Kumar added a comment - The memory shown at the guest os level is not correct as the guest OS is not aware of the memory optimization at the hypervisor level. In order to verify the actual allocation vmware counters should be observed. In this case as mentioned total memory > actual memory available. which means overcommit is working. this is not a bug.
        Hide
        prashant kumar mishra added a comment -

        closing it ,since its not an issue

        Show
        prashant kumar mishra added a comment - closing it ,since its not an issue

          People

          • Assignee:
            Bharat Kumar
            Reporter:
            prashant kumar mishra
          • Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

            • Created:
              Updated:
              Resolved:

              Development