You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@cloudstack.apache.org by Andrija Panic <an...@gmail.com> on 2013/10/24 22:53:01 UTC
Guest reports wrong amount of RAM with RAM overprovisioning
Hi,
I have updated to CS 4.2 (not sure that my problem has to do anything with
that...)
I have RAM overprovisioning level of 2 on Cluster level.
And after I start VM with compute ofering of 2GHZ (1x2gHz) and 2GB of ram,
it does invoke qemu-kvm with good parameters (2048M), but inside my VM when
doing top, or free -m, it reports rougly 50% of that amount:
Here is the ps axu | grep vname #from host on which VM is running:
root 17264 5.9 2.1 3447568 1090636 ? Sl 22:35 0:23
/usr/libexec/qemu-kvm -name i-2-248-VM -S -M rhel6.4.0 -enable-kvm *-m 2048
-*smp 1,sockets=1,cores=1,threads=1 -uuid 45087776-1238-4255-a87f-7ae6bcddc4a3
-nodefconfig -nodefaults -chardev socket,id=charmonitor,path=/
var/lib/libvirt/qemu/i-2-248-VM.monitor,server,nowait -mon
chardev=charmonitor,id=monitor,mode=control -rtc base=utc -no-shutdown
-device piix3-usb-uhci,id=usb,bus=pci.0,addr=0x1.0x2 -drive
file=/var/lib/libvirt/images/aa0d036a-f814-4cd8-991f-
d0a0ad21a7d4,if=none,id=drive-virtio-disk0,format=qcow2,cache=none -device
virtio-blk-pci,scsi=off,bus=pci.0,addr=0x4,drive=drive-
virtio-disk0,id=virtio-disk0,bootindex=2 -drive
if=none,media=cdrom,id=drive-ide0-1-0,readonly=on,format=raw,cache=none
-device ide-drive,bus=ide.1,unit=0,drive=drive-ide0-1-0,id=ide0-1-0,bootindex=1
-netdev tap,fd=23,id=hostnet0,vhost=on,vhostfd=36 -device
virtio-net-pci,netdev=hostnet0,id=net0,mac=06:b0:24:00:00:66,bus=pci.0,addr=0x3
-chardev pty,id=charserial0 -device isa-serial,chardev=charserial0,id=serial0
-device usb-tablet,id=input0 -vnc0.0.0.0:12,password -vga cirrus -device
virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x5
>From inside VM:
# free -m
*total* used free shared buffers cached
Mem: *853* 92 761 0 5 28
-/+ buffers/cache: 58 795
Swap: 4095 0 4095
Same for top.
Not sure what is happening, has to do something with RAM overprovisioning
factor ?
How is this possible, when qemu-kvm parameteres seems fine (-m 2048)
Thanks,
--
Andrija Panić
Re: Guest reports wrong amount of RAM with RAM overprovisioning
Posted by Marty Sweet <ms...@gmail.com>.
Thanks Nux, I mean the problem of Memory Balloon with KVM. I am currently
running CS4.2.0 (on Ubuntu 12.04.03 - Linux 3.5.0-34-generic) without these
issues.
Marty
On Sat, Feb 8, 2014 at 5:27 PM, Nux! <nu...@li.nux.ro> wrote:
> On 08.02.2014 17:02, Marty Sweet wrote:
>
>> Just wondering, in what version of CS was this issue introduced?
>>
>> Marty
>>
>
> It's in 4.3 (currently in development), some RPMs here if you feel like
> experimenting:
> http://tmp.nux.ro/cloudsnap430_4335/
>
> You need this in the /etc/cloudstack/agent/agent.properties:
> vm.memballoon.disable=true
>
>
> --
> Sent from the Delta quadrant using Borg technology!
>
> Nux!
> www.nux.ro
>
Re: Guest reports wrong amount of RAM with RAM overprovisioning
Posted by Nux! <nu...@li.nux.ro>.
On 08.02.2014 17:02, Marty Sweet wrote:
> Just wondering, in what version of CS was this issue introduced?
>
> Marty
It's in 4.3 (currently in development), some RPMs here if you feel like
experimenting:
http://tmp.nux.ro/cloudsnap430_4335/
You need this in the /etc/cloudstack/agent/agent.properties:
vm.memballoon.disable=true
--
Sent from the Delta quadrant using Borg technology!
Nux!
www.nux.ro
Re: Guest reports wrong amount of RAM with RAM overprovisioning
Posted by Marty Sweet <ms...@gmail.com>.
Just wondering, in what version of CS was this issue introduced?
Marty
On Fri, Feb 7, 2014 at 11:02 AM, Nux! <nu...@li.nux.ro> wrote:
> On 07.02.2014 10:58, Andrija Panic wrote:
>
>> Great, thx for that info, seems a little bit like cheating :) , but is
>> needed in some scenarios...
>>
>
> Well, depends. In some circumstances KSM can make a big difference and you
> certainly can't sell VMs with 2 GB ram and not have your customers
> complaining they only got half of that. :)
>
>
>
> --
> Sent from the Delta quadrant using Borg technology!
>
> Nux!
> www.nux.ro
>
Re: Guest reports wrong amount of RAM with RAM overprovisioning
Posted by Nux! <nu...@li.nux.ro>.
On 07.02.2014 10:58, Andrija Panic wrote:
> Great, thx for that info, seems a little bit like cheating :) , but is
> needed in some scenarios...
Well, depends. In some circumstances KSM can make a big difference and
you certainly can't sell VMs with 2 GB ram and not have your customers
complaining they only got half of that. :)
--
Sent from the Delta quadrant using Borg technology!
Nux!
www.nux.ro
Re: Guest reports wrong amount of RAM with RAM overprovisioning
Posted by Andrija Panic <an...@gmail.com>.
Great, thx for that info, seems a little bit like cheating :) , but is
needed in some scenarios...
Thanks
On 7 February 2014 11:54, Nux! <nu...@li.nux.ro> wrote:
> On 07.02.2014 10:44, Nux! wrote:
>
>> On 07.02.2014 08:43, Andrija Panic wrote:
>>
>>> hm, no, I have gave up on RAM overprovisioning, since I got no good
>>> response from any of the gurus here.
>>> If you find a solution, I would greatly appriciate it...
>>>
>>
>> Andrija,
>>
>> This is a known issue and it's how it works with other hypervisors,
>> too. This feature was not meant to be used in a public cloud (eg
>> selling VPSes) for obvious reasons, imo.
>> Starting with 4.3, thanks to Marcus Sorensen who wrote the patch,
>> we'll be able to ser overprovisioning AND not the the RAM shown inside
>> the guests messed with; basically it will rely on KSM for memory
>> deduplication. My advice is to start testing 4.3.
>>
>
> For more info read http://mail-archives.apache.
> org/mod_mbox/cloudstack-dev/201401.mbox/%3CCALFpzo7yx+N=RrbfUkwn7PHbf+Ra=
> 6880YQtJYGJXd2yjt_r1A@mail.gmail.com%3E
>
>
> --
> Sent from the Delta quadrant using Borg technology!
>
> Nux!
> www.nux.ro
>
--
Andrija Panić
--------------------------------------
http://admintweets.com
--------------------------------------
Re: Guest reports wrong amount of RAM with RAM overprovisioning
Posted by Nux! <nu...@li.nux.ro>.
On 07.02.2014 10:44, Nux! wrote:
> On 07.02.2014 08:43, Andrija Panic wrote:
>> hm, no, I have gave up on RAM overprovisioning, since I got no good
>> response from any of the gurus here.
>> If you find a solution, I would greatly appriciate it...
>
> Andrija,
>
> This is a known issue and it's how it works with other hypervisors,
> too. This feature was not meant to be used in a public cloud (eg
> selling VPSes) for obvious reasons, imo.
> Starting with 4.3, thanks to Marcus Sorensen who wrote the patch,
> we'll be able to ser overprovisioning AND not the the RAM shown inside
> the guests messed with; basically it will rely on KSM for memory
> deduplication. My advice is to start testing 4.3.
For more info read
http://mail-archives.apache.org/mod_mbox/cloudstack-dev/201401.mbox/%3CCALFpzo7yx+N=RrbfUkwn7PHbf+Ra=6880YQtJYGJXd2yjt_r1A@mail.gmail.com%3E
--
Sent from the Delta quadrant using Borg technology!
Nux!
www.nux.ro
Re: Guest reports wrong amount of RAM with RAM overprovisioning
Posted by Nux! <nu...@li.nux.ro>.
On 07.02.2014 08:43, Andrija Panic wrote:
> hm, no, I have gave up on RAM overprovisioning, since I got no good
> response from any of the gurus here.
> If you find a solution, I would greatly appriciate it...
Andrija,
This is a known issue and it's how it works with other hypervisors,
too. This feature was not meant to be used in a public cloud (eg selling
VPSes) for obvious reasons, imo.
Starting with 4.3, thanks to Marcus Sorensen who wrote the patch, we'll
be able to ser overprovisioning AND not the the RAM shown inside the
guests messed with; basically it will rely on KSM for memory
deduplication. My advice is to start testing 4.3.
--
Sent from the Delta quadrant using Borg technology!
Nux!
www.nux.ro
Re: Guest reports wrong amount of RAM with RAM overprovisioning
Posted by Andrija Panic <an...@gmail.com>.
hm, no, I have gave up on RAM overprovisioning, since I got no good
response from any of the gurus here.
If you find a solution, I would greatly appriciate it...
Thanks,
On 6 February 2014 21:42, motty cruz <mo...@gmail.com> wrote:
> Hi Andrija, did you resolve issue you were having with RAM and
> over-provision? I'm having same issues. my cluster is a KVM, I have
> CONFIG_VIRTIO_BALLOON=y
> in /boot/config-'uname -r'
>
> Thanks
>
>
> On Thu, Oct 24, 2013 at 2:03 PM, Andrija Panic <andrija.panic@gmail.com
> >wrote:
>
> > Just lowered overprovisioning of RAM from 2 to 1, and started the same VM
> > again, and I got good amount of RAM from inside the guest.
> >
> > when I do the ps aux thing from host, I get 100% same qemu-kvm
> > parameters, confirmed by diff...
> >
> > How is this possible ?
> >
> >
> > On 24 October 2013 22:53, Andrija Panic <an...@gmail.com> wrote:
> >
> > > Hi,
> > >
> > > I have updated to CS 4.2 (not sure that my problem has to do anything
> > with
> > > that...)
> > > I have RAM overprovisioning level of 2 on Cluster level.
> > >
> > > And after I start VM with compute ofering of 2GHZ (1x2gHz) and 2GB of
> > ram,
> > > it does invoke qemu-kvm with good parameters (2048M), but inside my VM
> > when
> > > doing top, or free -m, it reports rougly 50% of that amount:
> > >
> > > Here is the ps axu | grep vname #from host on which VM is running:
> > >
> > > root 17264 5.9 2.1 3447568 1090636 ? Sl 22:35 0:23
> > > /usr/libexec/qemu-kvm -name i-2-248-VM -S -M rhel6.4.0 -enable-kvm *-m
> > > 2048 -*smp 1,sockets=1,cores=1,threads=1 -uuid
> > 45087776-1238-4255-a87f-7ae6bcddc4a3
> > > -nodefconfig -nodefaults -chardev socket,id=charmonitor,path=/
> > > var/lib/libvirt/qemu/i-2-248-VM.monitor,server,nowait -mon
> > > chardev=charmonitor,id=monitor,mode=control -rtc base=utc -no-shutdown
> > > -device piix3-usb-uhci,id=usb,bus=pci.0,addr=0x1.0x2 -drive
> > > file=/var/lib/libvirt/images/aa0d036a-f814-4cd8-991f-
> > > d0a0ad21a7d4,if=none,id=drive-virtio-disk0,format=qcow2,cache=none
> > > -device virtio-blk-pci,scsi=off,bus=pci.0,addr=0x4,drive=drive-
> > > virtio-disk0,id=virtio-disk0,bootindex=2 -drive
> > > if=none,media=cdrom,id=drive-ide0-1-0,readonly=on,format=raw,cache=none
> > > -device
> > ide-drive,bus=ide.1,unit=0,drive=drive-ide0-1-0,id=ide0-1-0,bootindex=1
> > > -netdev tap,fd=23,id=hostnet0,vhost=on,vhostfd=36 -device
> > >
> >
> virtio-net-pci,netdev=hostnet0,id=net0,mac=06:b0:24:00:00:66,bus=pci.0,addr=0x3
> > > -chardev pty,id=charserial0 -device
> > isa-serial,chardev=charserial0,id=serial0
> > > -device usb-tablet,id=input0 -vnc0.0.0.0:12,password -vga cirrus
> -device
> > > virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x5
> > >
> > > From inside VM:
> > >
> > > # free -m
> > > *total* used free shared buffers
> > > cached
> > > Mem: *853* 92 761 0 5
> > > 28
> > > -/+ buffers/cache: 58 795
> > > Swap: 4095 0 4095
> > >
> > > Same for top.
> > > Not sure what is happening, has to do something with RAM
> overprovisioning
> > > factor ?
> > > How is this possible, when qemu-kvm parameteres seems fine (-m 2048)
> > >
> > > Thanks,
> > > --
> > >
> > > Andrija Panić
> > >
> >
> >
> >
> > --
> >
> > Andrija Panić
> > --------------------------------------
> > http://admintweets.com
> > --------------------------------------
> >
>
--
Andrija Panić
--------------------------------------
http://admintweets.com
--------------------------------------
Re: Guest reports wrong amount of RAM with RAM overprovisioning
Posted by motty cruz <mo...@gmail.com>.
Hi Andrija, did you resolve issue you were having with RAM and
over-provision? I'm having same issues. my cluster is a KVM, I have
CONFIG_VIRTIO_BALLOON=y
in /boot/config-'uname -r'
Thanks
On Thu, Oct 24, 2013 at 2:03 PM, Andrija Panic <an...@gmail.com>wrote:
> Just lowered overprovisioning of RAM from 2 to 1, and started the same VM
> again, and I got good amount of RAM from inside the guest.
>
> when I do the ps aux thing from host, I get 100% same qemu-kvm
> parameters, confirmed by diff...
>
> How is this possible ?
>
>
> On 24 October 2013 22:53, Andrija Panic <an...@gmail.com> wrote:
>
> > Hi,
> >
> > I have updated to CS 4.2 (not sure that my problem has to do anything
> with
> > that...)
> > I have RAM overprovisioning level of 2 on Cluster level.
> >
> > And after I start VM with compute ofering of 2GHZ (1x2gHz) and 2GB of
> ram,
> > it does invoke qemu-kvm with good parameters (2048M), but inside my VM
> when
> > doing top, or free -m, it reports rougly 50% of that amount:
> >
> > Here is the ps axu | grep vname #from host on which VM is running:
> >
> > root 17264 5.9 2.1 3447568 1090636 ? Sl 22:35 0:23
> > /usr/libexec/qemu-kvm -name i-2-248-VM -S -M rhel6.4.0 -enable-kvm *-m
> > 2048 -*smp 1,sockets=1,cores=1,threads=1 -uuid
> 45087776-1238-4255-a87f-7ae6bcddc4a3
> > -nodefconfig -nodefaults -chardev socket,id=charmonitor,path=/
> > var/lib/libvirt/qemu/i-2-248-VM.monitor,server,nowait -mon
> > chardev=charmonitor,id=monitor,mode=control -rtc base=utc -no-shutdown
> > -device piix3-usb-uhci,id=usb,bus=pci.0,addr=0x1.0x2 -drive
> > file=/var/lib/libvirt/images/aa0d036a-f814-4cd8-991f-
> > d0a0ad21a7d4,if=none,id=drive-virtio-disk0,format=qcow2,cache=none
> > -device virtio-blk-pci,scsi=off,bus=pci.0,addr=0x4,drive=drive-
> > virtio-disk0,id=virtio-disk0,bootindex=2 -drive
> > if=none,media=cdrom,id=drive-ide0-1-0,readonly=on,format=raw,cache=none
> > -device
> ide-drive,bus=ide.1,unit=0,drive=drive-ide0-1-0,id=ide0-1-0,bootindex=1
> > -netdev tap,fd=23,id=hostnet0,vhost=on,vhostfd=36 -device
> >
> virtio-net-pci,netdev=hostnet0,id=net0,mac=06:b0:24:00:00:66,bus=pci.0,addr=0x3
> > -chardev pty,id=charserial0 -device
> isa-serial,chardev=charserial0,id=serial0
> > -device usb-tablet,id=input0 -vnc0.0.0.0:12,password -vga cirrus -device
> > virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x5
> >
> > From inside VM:
> >
> > # free -m
> > *total* used free shared buffers
> > cached
> > Mem: *853* 92 761 0 5
> > 28
> > -/+ buffers/cache: 58 795
> > Swap: 4095 0 4095
> >
> > Same for top.
> > Not sure what is happening, has to do something with RAM overprovisioning
> > factor ?
> > How is this possible, when qemu-kvm parameteres seems fine (-m 2048)
> >
> > Thanks,
> > --
> >
> > Andrija Panić
> >
>
>
>
> --
>
> Andrija Panić
> --------------------------------------
> http://admintweets.com
> --------------------------------------
>
Re: Guest reports wrong amount of RAM with RAM overprovisioning
Posted by Andrija Panic <an...@gmail.com>.
Just lowered overprovisioning of RAM from 2 to 1, and started the same VM
again, and I got good amount of RAM from inside the guest.
when I do the ps aux thing from host, I get 100% same qemu-kvm
parameters, confirmed by diff...
How is this possible ?
On 24 October 2013 22:53, Andrija Panic <an...@gmail.com> wrote:
> Hi,
>
> I have updated to CS 4.2 (not sure that my problem has to do anything with
> that...)
> I have RAM overprovisioning level of 2 on Cluster level.
>
> And after I start VM with compute ofering of 2GHZ (1x2gHz) and 2GB of ram,
> it does invoke qemu-kvm with good parameters (2048M), but inside my VM when
> doing top, or free -m, it reports rougly 50% of that amount:
>
> Here is the ps axu | grep vname #from host on which VM is running:
>
> root 17264 5.9 2.1 3447568 1090636 ? Sl 22:35 0:23
> /usr/libexec/qemu-kvm -name i-2-248-VM -S -M rhel6.4.0 -enable-kvm *-m
> 2048 -*smp 1,sockets=1,cores=1,threads=1 -uuid 45087776-1238-4255-a87f-7ae6bcddc4a3
> -nodefconfig -nodefaults -chardev socket,id=charmonitor,path=/
> var/lib/libvirt/qemu/i-2-248-VM.monitor,server,nowait -mon
> chardev=charmonitor,id=monitor,mode=control -rtc base=utc -no-shutdown
> -device piix3-usb-uhci,id=usb,bus=pci.0,addr=0x1.0x2 -drive
> file=/var/lib/libvirt/images/aa0d036a-f814-4cd8-991f-
> d0a0ad21a7d4,if=none,id=drive-virtio-disk0,format=qcow2,cache=none
> -device virtio-blk-pci,scsi=off,bus=pci.0,addr=0x4,drive=drive-
> virtio-disk0,id=virtio-disk0,bootindex=2 -drive
> if=none,media=cdrom,id=drive-ide0-1-0,readonly=on,format=raw,cache=none
> -device ide-drive,bus=ide.1,unit=0,drive=drive-ide0-1-0,id=ide0-1-0,bootindex=1
> -netdev tap,fd=23,id=hostnet0,vhost=on,vhostfd=36 -device
> virtio-net-pci,netdev=hostnet0,id=net0,mac=06:b0:24:00:00:66,bus=pci.0,addr=0x3
> -chardev pty,id=charserial0 -device isa-serial,chardev=charserial0,id=serial0
> -device usb-tablet,id=input0 -vnc0.0.0.0:12,password -vga cirrus -device
> virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x5
>
> From inside VM:
>
> # free -m
> *total* used free shared buffers
> cached
> Mem: *853* 92 761 0 5
> 28
> -/+ buffers/cache: 58 795
> Swap: 4095 0 4095
>
> Same for top.
> Not sure what is happening, has to do something with RAM overprovisioning
> factor ?
> How is this possible, when qemu-kvm parameteres seems fine (-m 2048)
>
> Thanks,
> --
>
> Andrija Panić
>
--
Andrija Panić
--------------------------------------
http://admintweets.com
--------------------------------------