CPU performance test results in three scenarios
by Tommy Sway
Hello, I recently conducted a test to compare the CPU performance of a
physical machine, a high performance VM (pass-through CPU, PIN CPU), and a
common VM.
Here are the results:
1. Physical machine:
# sysbench --test=cpu --cpu-max-prime=100000000 --threads=2 run
WARNING: the --test option is deprecated. You can pass a script name or path
on the command line without any options.
sysbench 1.0.17 (using system LuaJIT 2.0.4)
Running the test with following options:
Number of threads: 2
Initializing random number generator from current time
Prime numbers limit: 100000000
Initializing worker threads...
Threads started!
CPU speed:
events per second: 0.01
General statistics:
total time: 346.0588s
total number of events: 2
Latency (ms):
min: 345943.52
avg: 346001.09
max: 346058.65
95th percentile: 100000.00
sum: 692002.17
Threads fairness:
events (avg/stddev): 1.0000/0.00
execution time (avg/stddev): 346.0011/0.06
2. High Performance VM:
# sysbench --test=cpu --cpu-max-prime=100000000 --threads=2 run
WARNING: the --test option is deprecated. You can pass a script name or path
on the command line without any options.
sysbench 1.0.17 (using system LuaJIT 2.0.4)
Running the test with following options:
Number of threads: 2
Initializing random number generator from current time
Prime numbers limit: 100000000
Initializing worker threads...
Threads started!
CPU speed:
events per second: 0.01
General statistics:
total time: 351.8625s
total number of events: 2
Latency (ms):
min: 351416.59
avg: 351639.01
max: 351861.43
95th percentile: 100000.00
sum: 703278.02
Threads fairness:
events (avg/stddev): 1.0000/0.00
execution time (avg/stddev): 351.6390/0.22
3. Common VM:
# sysbench --test=cpu --cpu-max-prime=100000000 --threads=2 run
WARNING: the --test option is deprecated. You can pass a script name or path
on the command line without any options.
sysbench 1.0.17 (using system LuaJIT 2.0.4)
Running the test with following options:
Number of threads: 2
Initializing random number generator from current time
Prime numbers limit: 100000000
Initializing worker threads...
Threads started!
CPU speed:
events per second: 0.01
General statistics:
total time: 354.9108s
total number of events: 2
Latency (ms):
min: 354761.26
avg: 354835.99
max: 354910.73
95th percentile: 100000.00
sum: 709671.99
Threads fairness:
events (avg/stddev): 1.0000/0.00
execution time (avg/stddev): 354.8360/0.07
Result:
There is little difference in CPU performance between these three scenarios,
so why distinguish between high performance and regular virtual machines? I
don't think it makes much sense.
3 years, 7 months
The provided authorization grant for the auth code has expired.
by Nicolás
Hi,
A time ago I posted a similar question and I couldn't get it solved. I
couldn't spend more time on this up until now, so I'm trying again and
having the same error, which I'm unable to fix. I would really
appreciate some help since I cannot find any documentation for this.
I'm restoring a full ovirt engine backup, having used the --scope=all
option, for oVirt 4.3.
I restored the backup on a fresh CentOS7 machine. The process went well,
but when trying to log into the restored authentication system I get the
following message which won't allow me to log in:
The provided authorization grant for the auth code has expired.
What does that mean and how can it be fixed?
I'm also attaching a screenshot.
Thanks.
Nicolás
3 years, 7 months
Question about cloud-init
by tommy
Hello, I have a question about cloud-init:
After installing vm, I installed cloud-init immediately, and enabled its
four services, and then I can use cloud-init in the admin interface.
However, I don't know how cloud-init knows where to get the relevant
information (such as host name, SSH secret key, etc.) that I configured on
the admin interface ? For I haven't configured the data source for
cloud-init.
Why it can works ?
Thanks!
3 years, 7 months
Question about pci pass-thru
by Tony Pearce
I have recently added a fresh installed host on 4.4, with 3 x nvidia gpu's
which have been passed through to a guest VM instance. The host also has a
1TB nvme SSD drive which we are using for the host OS install and /var/
partition size was modified to allow maximum space for the guest, the
intent is to give maximum space to a guest for fast SSD storage. Because
this is a partition, I have about 700GB that can be added or mapped to a
guest "disk" image that is locked to the host.
After doing the above and after learning how simple it was to map the three
GPUs in pass-through mode, my question is - can I simply map a PCI SSD
through to a guest in the same fashion as the GPUs?
I have not yet tried and do not find much results via google search. The
intent on this would be to provision the whole SSD to the guest (instead of
using the SSD for the host OS and some of that space for a guest (disk)
image. In this case I would use another disk for the host OS but that is
fine.
Has anyone done such a thing (PCI SSD passthrough) and are there any
concerns? (apart from the normal concerns with VM and host passthrough)
Kind regards,
Tony Pearce
3 years, 7 months
Set fixed VNC/Spice Password for VMs.
by Merlin Timm
Good day to all,
I have a question about the console configuration of the VMs:
By default, for each console connection to a VM, a password is set for
120 seconds, after that you can't use it again. We currently have the
following concern:
We want to access and control the VMs via the VNC/Spice of the Ovirt
host. We have already tried to use the password from the console.vv for
the connection and that works so far. Unfortunately we have to do this
every 2 minutes when we want to connect again. We are currently building
an automatic test pipeline and for this we need to access the VMs
remotely before OS start and we want to be independent of a VNC server
on the guest. This is only possible if we could connect to the VNC/Spice
server from the Ovirt host.
My question: would it be possible to fix the password or read it out via
api every time you want to connect?
I would appreciate a reply very much!
Best regards
Merlin Timm
3 years, 7 months
Happy SysAdmin Day!
by Strahil Nikolov
Happy SysAdmin Day!
I want to thank Duck and wish to all fellow SysAdmins in the oVirt community a Happy Holliday !
Best Regards,Strahil Nikolov
3 years, 7 months
adding new host failed with error "CPU does not match the Cluster CPU Type lscpu"
by jimod4＠yahoo.com
i get the below error when i try to add a new kvm host to the cluster. I know that the cpu is a haswell. if i run "virsh capabilities" it returns
<model>Haswell-noTSX-IBRS</model>. My KVM server is redhat 8.4 fully patched today on an DL-360 Gen9. oVirt is at 4.4.7.7-1.el8. i have tried many diffent CPU settings for the cluster and none of them work. this is new oVirt install.
ERROR:
"The host CPU does not match the Cluster CPU Type and is running in a degraded mode. It is missing the following CPU flags: vmx, model_Haswell-noTSX, nx. Please update the host CPU microcode or change the Cluster CPU Type."
MIRCOCODE:
root@kvm09 ~]# dmesg | grep 'microcode'
[ 0.000000] microcode: microcode updated early to revision 0x46, date = 2021-01-27
[ 1.625108] microcode: sig=0x306f2, pf=0x1, revision=0x46
[ 1.625753] microcode: Microcode Update Driver: v2.2.
3 years, 7 months
Change lease storage domain while VM running
by Gianluca Cecchi
Hello,
on a 4.4.6 environment where I'm moving storage I want to change the lease
domain for an HA VM.
The VM is running and if I go and edit, change sd and then press ok button,
I get a window with:
"
Changes that can be applied immediately:
VM Lease
"
I don't select the "apply later" check box. But then instead when I
confirm, I get an error window with:
"
vm_name:
Cannot edit VM. Hot swapping of VM leases is not supported, please
remove the current lease and then set a new one.
"
Possibly it is possible to hot add a lease but not swap? So I think the
first window should be changed. Is this a known issue?
No problem if I change the lease storage domain while the VM is powered off.
I don't know if it is already fixed in 4.4.7... I'm going to check next
days..
Gianluca
3 years, 7 months
Re: Upload Image to NFS Storage Domain Doesn't Work
by Eyal Shenitzky
+Bella Khizgiyaev <bkhizgiy(a)redhat.com> can you please take a look?
On Fri, 30 Jul 2021 at 16:15, Nur Imam Febrianto <nur_imam(a)outlook.com>
wrote:
> Hi,
>
>
>
> Lastly in 4.4.6, I have a problem that I couldn’t upload any image to
> storage domain without clicking Test Connection. Well, after it was
> acknowledge as a bug and solved at 4.4.7 now I have similar issue. If I
> upload any image to Storage Domain that uses NFS it only show locked, and
> then showing OK even if no data are uploaded (can confirm 5GB ISO showing
> OK after only 1 second). This issue only occurs with NFS backend, another
> storage domain that uses iSCSI work fine though. Any idea how to solve this
> ? I’m using NFS Storage Domain for storing ISO and this issue bugged me all
> the time.
>
>
>
> Thanks before.
>
>
>
> Regards,
>
>
>
> Nur Imam Febrianto
>
>
>
>
> _______________________________________________
> Users mailing list -- users(a)ovirt.org
> To unsubscribe send an email to users-leave(a)ovirt.org
> Privacy Statement: https://www.ovirt.org/privacy-policy.html
> oVirt Code of Conduct:
> https://www.ovirt.org/community/about/community-guidelines/
> List Archives:
> https://lists.ovirt.org/archives/list/users@ovirt.org/message/ACUKBWAM2OR...
>
--
Regards,
Eyal Shenitzky
3 years, 7 months
