[one-users] VM in boot state for a very long time

[Liu, Guang Jun (Gene)]

Hi Javier,

Thank you for your reply!
After read this blog http://blog.opennebula.org/?p=3473 I started to 
concern some settings of my system. The test (a demo/show for my 
testing organization management team) I performed was that 20 VMs were 
initiated as a batch. We have 4 hosts as computing nodes. We use linux 
bridge network w/ external DHCP server. (we chose network mode as 
"default")

I use all the default settings from a fresh new installation. I tried 
to tune some of the parameters in oned.conf and sched.conf as below --

oned.conf:
MANAGER_TIMER           = 30
MONITORING_INTERVAL     = 60
MONITORING_THREADS      = 50

IM_MAD = [
      name       = "kvm",
      executable = "one_im_ssh",
      arguments  = "-r 0 -t 40 kvm" ]

VM_MAD = [
    name       = "kvm",
    executable = "one_vmm_exec",
    arguments  = "-t 40 -r 0 kvm",
    default    = "vmm_exec/vmm_exec_kvm.conf",
    type       = "kvm" ]

TM_MAD = [
    executable = "one_tm",
    arguments  = "-t 200 -d dummy,lvm,shared,qcow2,ssh,vmfs,iscsi,ceph" 
]

DATASTORE_MAD = [
    executable = "one_datastore",
    arguments  = "-t 200 -d dummy,fs,vmfs,iscsi,lvm,ceph"
]

HM_MAD = [
    executable = "one_hm" ]

sched.conf:
SCHED_INTERVAL = 180

MAX_VM       = 300
MAX_DISPATCH = 200
MAX_HOST     = 1

LIVE_RESCHEDS  = 0

HYPERVISOR_MEM = 0.1

DEFAULT_SCHED = [
    policy = 0,
    RANK = RUNNING_VMS
]

Regards,

Gene Liu

On Wed 11 Dec 2013 08:54:31 AM EST, Javier Fontan wrote:
> This can be caused by several things.
>
> * There are already 15 actions being done by the VMM. By default the
> drivers are only allowed to execute 15 actions at the same time and
> the next actions are enqueued. Have you launched a batch of actions to
> lots of VMs?
>
> * There is already an action being executed in that host that is
> hanging. The drivers only execute one action per Host at the same time
> as sometimes libvirt does not play well with several actions. Can you
> check that there is no hanging action script executing in that host? A
> "ps" and check for script executions from /var/tmp/one is a good way
> to check for them.
>
> * The virtual network drivers are taking too much time. What are the
> network drivers you are using?
>
> Cheers
>
> On Tue, Dec 10, 2013 at 5:54 PM, Liu, Guang Jun (Gene)
> <Gene.Liu at alcatel-lucent.com> wrote:
>> Hi there,
>>
>> I am running 4.2 (RHEL6.4, KVM). I experienced that VM stays in "BOOT"
>> state for a very long time (~15 minutes). The image copy is fast (VM
>> state goes to BOOT from prolog). There is not any log and the servers
>> (both the computing node/host, and one front-server) are idle. Can
>> anyone suggest me where I should look at to figure out the cause of VM
>> in "BOOT" so long?
>>
>> Thanks very much!
>>
>> --
>> Gene Liu
>>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opennebula.org
>> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>
>
>